APBRmetrics

Dan Rosenbaum



Joined: 03 Jan 2005
Posts: 541
Location: Greensboro, North Carolina

PostPosted: Fri Feb 04, 2005 1:13 pm Post subject: Shouldn't possession usage count all potential assists? Reply with quote
Indivdual Possession Usage

When we calculate individual possession usage (and offensive ratings), we count field goal attempts, free throw attempts, assists, turnovers and (sometimes) offensive rebounds.

So when the PG comes down the floor and passes to the SG cutting to the basket who then shoots the ball, that pass increases possession usage if (a) the shot goes in, but not if (b) the shot is missed (even if there is a foul that leads to free throws).

In both cases (a) and (b), the PG generated a shot attempt of exactly the same quality. But it only counts as using a possession if the shot happens to go in the basket. Had the PG shot the ball himself, it would count towards possession usage, regardless of whether the shot went in. But when he passes it off to a teammate, we only count it if the shot goes in.

That seems fundamentally inconsistent, but I doubt that I am the first to realize that. I think the reason we generally only count assists and not all potential assists is that we don't know the number of potential assists. But I think we can estimate the number of potential assists.

potential assists = assists * (0.5 * points / field goals made) / teammates' true shooting percentage

The second term (0.5 * points / field goals made) scales up the assists to include free throws and three pointers. Since it takes more passes to generate an assist on a three pointer, it is only fair to allow the assist to count more. It would be better with this formula to take out the free throw points due to non-shooting fouls. (Perhaps this could be done in some sort of average sense using the data that Ed provided for us.)

The third term in essence counts the potential assists where the shot was missed. A true shooting percentage of 50 percent would imply that it takes two potential assists to generate an assist. Ideally this term would use the teammates' true shooting percentage on attempts after potential assists, although presently we do not have the data to calculate that.

This next paragraph has been edited to fix mistakes and add clarity.

DeanO uses a one half multiplier, but only counts assists, not potential assists. This formulation uses a smaller multiplier of one third, but counts all potential assists. This formulation likely will result in assists counting a little higher in my formulation than in DeanO's. Compared to DeanO's one half multiplier, this formulation probably ends up with something between 0.53 and 0.61 with the higher multiplier for players on poor shooting teams. (Those players will have higher possession usage, but lower offensive ratings - kind of like what happens with them with shooting.)

Individual Offensive Rating

When computing offensive ratings, we are typically measuring the points created per possession. The assist possessions go into the denominator and (if I remember correctly) the assist possessions multiplied by two (or something a little greater than two) go into the numerator. Thus, getting assists will invariably push a player towards an offensive rating of 200 (or higher) with higher turnovers due to errant assist attempts being the only the only thing holding back high assisters from having really high offensive ratings.

My formulation suggests another way for assists to enter offensive ratings. In the denominator they would enter in the same way as they enter possession usage. In the numerator the number of possessions would be multiplied by 2*true shooting percentage (for potential assists). Thus, potential assists would come into the formula at the rate of points per true shot attempt (for potential assists) for the rest of the team.

Thus, a high percentage shooting PG would decrease his offensive rating by giving lots of assists to his low percentage shooting teammates. His possession usage would go up, but his offensive rating would go down.

This is why I have been probing folks about the true shooting percentage on potential assists vs. non-potential assists. If, like Ed calculated in a short study last season, the true shooting percentage for potential assists is about 10 percentage points higher than that for non-potential assists, then assists will tend to raise offensive ratings quite a bit. If not, they would not.

This is a pretty long post, so I will stop know. This post is challenging my ability to organize thoughts, so I hope it makes some sense. I may edit this original post if subsequent posts point out where I need to be clearer.

Last edited by Dan Rosenbaum on Sat Feb 05, 2005 1:48 pm; edited 2 times in total
Back to top
View user's profile Send private message Send e-mail Visit poster's website Yahoo Messenger
bchaikin



Joined: 27 Jan 2005
Posts: 690
Location: cleveland, ohio

PostPosted: Fri Feb 04, 2005 1:38 pm Post subject: Reply with quote
concerning "Individual Possession Usage", this is what i call Possession Factor, or touches per minute, and its calculated from the raw numbers...

you can find these touches/min numbers listed for all players over the past 27 years in the NBA stats database (or the superDB) at www.bballsports.com, and the instructions file to the application has a statistics definition file talking about it...

it is this number the simulation uses to determine how often each player should handle the ball...
Back to top
View user's profile Send private message Send e-mail Visit poster's website
Dan Rosenbaum



Joined: 03 Jan 2005
Posts: 541
Location: Greensboro, North Carolina

PostPosted: Fri Feb 04, 2005 1:45 pm Post subject: Reply with quote
bchaikin wrote:
concerning "Individual Possession Usage", this is what i call Possession Factor, or touches per minute, and its calculated from the raw numbers...

you can find these touches/min numbers listed for all players over the past 27 years in the NBA stats database (or the superDB) at www.bballsports.com, and the instructions file to the application has a statistics definition file talking about it...

it is this number the simulation uses to determine how often each player should handle the ball...

What do you mean by "it is calculated from the raw numbers"? To me, that means that over the past 27 years you have charted exactly how many times every player in the NBA has touched the ball. Is that what you have done? Or have you done some charting from a sample of games and used the patterns from those games to estimate the touches from traditional statistics?

By the way, what exactly is a touch? If a PG brings the ball down, passes to the SG, who passes back to the PG, who passes to the SF, who passes to the SG, who passes to the PG, who lobs the ball to the PF who dunks the ball, how are the touches counted in that possession?
Back to top
View user's profile Send private message Send e-mail Visit poster's website Yahoo Messenger
bchaikin



Joined: 27 Jan 2005
Posts: 690
Location: cleveland, ohio

PostPosted: Fri Feb 04, 2005 2:26 pm Post subject: Reply with quote
here are some APBR_analysis postings (numbers 883-886) from 4/2002, these should shed some light on your questions...

In APBR_analysis@y..., bchaikin@a... wrote:

i calculate possession factor (touches per min) by:

poss fact =
shots + passes + turnovers + times fouled = FGA + AST/X + TO +FTA/Y

where X and Y are calculated yearly and by team...

Are X and Y secret?

Let me also ask whether the factor matches what you would see in a
game. For instance, if we calculate a player's touches per minute to
be 2 per minute, would that mean that, if he played 35 minutes, he
would touch the ball, on average, 70 distinct times in those 35
minutes?

yes - but my original research/analysis took place over the span of 3 seasons in the late 80s / early 90s, watching just under 1000 NBA games on tape, charting team and individual player ball possessions. the formulas are extrapolated from this data and streamlined for each season and each team (based on avg team yearly possessions and then each team's calculated possessions)...

I ask because the formula seems to not include rebounds. Obviously,
guys touch the ball when they get rebounds. If X and Y are team and
year specific, then it is possible to have a couple guys with very
different rebounding totals with similar possession factors. For
instance, Jon Barry and Ben Wallace aren't terribly different in the
main factors you list. Wallace has lower totals in all but FTA. But
he had 800 more rebounds than Barry. How do these guys stack up?

think about it...to calculate how often a player handled the ball on offense you either have to count all the times he first gained possession of the ball while on offense, or what he did with the ball (how he got rid of it) each time he had the ball...again, what can a player do with the ball once he has it in his hands? he can either shoot it, pass it, get fouled, or turn it over (FGA + AST/X + FTA/Y + TO). other things can occur (off foul, double fouls, jump ball, techs, etc) but these occur only occassionally such that they can be ignored - if they were all measured they could be included...

or how can a player gain possession of the ball while on offense? he can either start the team possession (get a pass from a player from out of bounds after the opposition has scored), start the team possession by catching an inbounds pass on an out-of-bounds play, simply receive a pass from another player on the court, or get an offensive rebound. i do not use def rebs because rarely does a player score immediately after getting a def reb (players do occassionally get rebounds, dribble the length of the court, and score, but the total # of times is insignificant compared to the total possessions and this parameter is tough to measure. if we had these numbers they could be added)..

as for barry and wallace, i just uploaded the 01-02 reg seas stats to the online database so you can compare them there. just limit the database to the 01-02 season, and sort by possession factor, and you'll see how two players can have the same possession factor but different stats that when added up and divided by minutes played equal similar touches-per-minute ratings (poss fact)...

also you can have two players with absolutely identical stats from two different seasons and they could easily have two different possession factors - because part of the rating is based on league averages. a simple example is shot blocking - two players can have the same # of blocks and minutes played in two different seasons but have their shot block rating vary by as much as half a percent or more based on what the rest of the league did that specific season. example - this past season jermaine o'neal blocked 166 shots in 2707 minutes and had a shot blocking rating of 3.6% (he blocked 3.6 of every 100 shots taken by the pacers opposition). in 1978-79 rich kelley of the jazz blocked 166 shots in 2705 minutes but had a shot blocking rating of only 3.2%....

What was the original goal of the stat? Is it better to say that it
is an estimate of touches on the offensive end?

the goal was to determine exactly how often each player handled the ball on offense so the computer could model the game, i.e. knowing how often each player handled the ball on offense and what he did with it (how often he would shoot, pass, get fouled, TO) once he did get the ball makes for an easy model for the computer to simply play a game...

are the formulas "secret"? not really. complicated? not at all - four parameters (statistical categories - FGA, TO, passes, times fouled) and two variables. but there are upwards of 30+ formulas per season (one for each team), every season, although each is quite similar, and some modifictions for players traded between teams of widely varying game pace...

the best way to explain this would be to download a copy of the simulation software i developed (www.bballsports.com) - again its free - and when you run a full season for any team it charts ALL ball possessions such that they can be added up easily. this is the best way to see "possession factors" or "touches per minute ratings" in action. plus the online stats databases lists possession factors for every player since 1977-78 for the NBA and even earlier for players in the ABA...
Back to top
View user's profile Send private message Send e-mail Visit poster's website
Dan Rosenbaum



Joined: 03 Jan 2005
Posts: 541
Location: Greensboro, North Carolina

PostPosted: Fri Feb 04, 2005 2:56 pm Post subject: Reply with quote
Bob, thank you very much for finding these old posts. I am not sure that I had read these posts, but they were almost exactly what I suspected that you were doing. You are estimating touches based upon current traditional statistics based upon patterns derived from past charting. I think this is a great way to combine charting and traditional statistics.

I hate to hijack this thread away from my original topic, but do you worry that with the new defensive rules and dramatic increase in three pointers that patterns for touches that you developed more than a decade ago may not apply so well today?

Did you estimate X and Y separately by position in case the touch patterns differ by position?

I still am a little confused what a touch is conceptually? It surely is not how many times you touch the ball in a possession. Otherwise, I think you would have multipliers on FGAs and TOs as well. What in your mind are touches measuring? Should we think of it as some weighted average of FGA, TO, FTA, and AS?
Back to top
View user's profile Send private message Send e-mail Visit poster's website Yahoo Messenger
HoopStudies



Joined: 30 Dec 2004
Posts: 706
Location: Near Philadelphia, PA

PostPosted: Fri Feb 04, 2005 6:17 pm Post subject: Re: Shouldn't possession usage count all potential assists? Reply with quote
Dan Rosenbaum wrote:

If we continued to give the passer one third of the credit for a potential assist, this would result in assists generating more than twice as much possession usage. This would result in point guards rising to the top of possession usage charts, which I think makes sense since they tend to control the ball much more than other players.


I don't like to think about possessions this way, but touches, yes. PGs definitely touch the ball a ton, but the frequency with which they impact a final possession does tend to be small. Not universally small by any means. Magic Johnson was way up there. Marbury uses a lot. KJ did. Nash is up at 24% this year, behind Stoudemire.

I definitely think that twice as much "possession" usage would be wrong. 40% of a team's possessions being generated by a point guard just doesn't make sense with a lot of the offense being carried out through the other 4 guys.

But the point in the subject is correct -- it would be nice to have potential assists. The framework for individual possessions would be modified, including a term in the denominator and modifying the allocation of credit on it. (As I think I explained in the book, I had to assume about a 70-80% success rate on passes that would be assists. If we find something different, it can be modified.)

Dan Rosenbaum wrote:

Individual Offensive Rating

When computing offensive ratings, we are typically measuring the points created per possession. The assist possessions go into the denominator and (if I remember correctly) the assist possessions multiplied by two (or something a little greater than two) go into the numerator. Thus, getting assists will invariably push a player towards an offensive rating of 200 (or higher) with higher turnovers due to errant assist attempts being the only the only thing holding back high assisters from having really high offensive ratings.

My formulation suggests another way for assists to enter offensive ratings. In the denominator they would enter in the same way as they enter possession usage. In the numerator the number of possessions would be multiplied by 2*true shooting percentage (for potential assists). Thus, potential assists would come into the formula at the rate of points per true shot attempt (for potential assists) for the rest of the team.


Almost this thing is in my calculation of scoring possessions. I do account for effective field goal percentage of teammates, but not true shooting percentage (which includes foul shots). This has been a tough thing for me because I do believe that assists should be awarded for getting to the line. If they were to start doing so, some minor modifications to the theory would be done. The framework of individual possessions certainly allows this. It could be done without a change in scoring, too, and I have done it in the past. The relatively small error induced in my estimate of shots assisted on gets magnified a bit if you include foul shots. So, I remember Adrian Dantley, who went to the line a ton on shots that weren't going to be assists, lost a lot of credit that he shouldn't have lost.

Dan Rosenbaum wrote:

Thus, a high percentage shooting PG would decrease his offensive rating by giving lots of assists to his low percentage shooting teammates. His possession usage would go up, but his offensive rating would go down.


This is exactly what happens for Jason Kidd, for example. He is a poor shooting guard, which makes it all the more impressive that his teammates have shot reasonably well (not great).

This accounting was an important part of the method that I worked out in 1989, I remember. I was working a summer job at a national lab where the job didn't keep me all that busy, so I walked around with a basketball in my hand and worked on this theory. It makes the formulas much more complex, but it features the interaction that does happen in the game.
_________________
Dean Oliver
Author, Basketball on Paper
The postings are my own & don't necess represent positions, strategies or opinions of employers.
Back to top
View user's profile Send private message Visit poster's website
HoopStudies



Joined: 30 Dec 2004
Posts: 706
Location: Near Philadelphia, PA

PostPosted: Fri Feb 04, 2005 6:25 pm Post subject: Reply with quote
Dan Rosenbaum wrote:

I hate to hijack this thread away from my original topic, but do you worry that with the new defensive rules and dramatic increase in three pointers that patterns for touches that you developed more than a decade ago may not apply so well today?

Did you estimate X and Y separately by position in case the touch patterns differ by position?

I still am a little confused what a touch is conceptually? It surely is not how many times you touch the ball in a possession. Otherwise, I think you would have multipliers on FGAs and TOs as well. What in your mind are touches measuring? Should we think of it as some weighted average of FGA, TO, FTA, and AS?


Let me try to answer the question because I did some work to reproduce Bob's calcs a while ago. They do estimate actual ball touches in a possession, at least the ones in the front court. Dennis Rodman's immense number of defensive touches on defensive rebounds don't count.

Stuart McKibbin did a count that Bob duplicated quite accurately with his formula. I haven't seen other tests, but we can do it with Roland/Ed's charting.

The estimate of X and Y is somewhat team-specific, but using league-wide numbers doesn't make a big difference. Although I admit I don't know how to use the numbers to say what a difference is. The multiplier on assists is quite large and smaller on FTA. I never did sit down to think about why the estimate works or whether it could be done better.

I know this is a critical part of Bob's simulator. To me, it is just interesting. Very interesting. Bob's simulator and my individual win-loss records often end up similar through apparently very different methods.

The other thing I never sat down to think about is how his simulator partitions touches when players get blended together.
_________________
Dean Oliver
Author, Basketball on Paper
The postings are my own & don't necess represent positions, strategies or opinions of employers.
Back to top
View user's profile Send private message Visit poster's website
bchaikin



Joined: 27 Jan 2005
Posts: 690
Location: cleveland, ohio

PostPosted: Fri Feb 04, 2005 9:49 pm Post subject: Reply with quote
sorry, my above equation wasn't quite accurate, let me try again. to calculate a player's possessions (touches) i use this:

player possessions =

shots + turnovers + passes thrown + # times fouled =

FGA + TO + AST/X + FTA/Y

and to calculate his possession factor (touches/min) i use:

poss fact = (FGA + TO + AST/X + FTA/Y)/MIN

I still am a little confused what a touch is conceptually?

any time a player has the ball on offense and subsequently shoots, passes (my definition of a pass, see below), gets fouled, or turns the ball over...

It surely is not how many times you touch the ball in a possession. Otherwise, I think you would have multipliers on FGAs and TOs as well. What in your mind are touches measuring? Should we think of it as some weighted average of FGA, TO, FTA, and AS?

to calculate passes way back when i first did this, the only passes i counted as "passes" were those that in my opinion had to have had an opportunity to be an assist, since assists were the only statistical parameter i had access to that correlated in some way with passes (i.e. assists are a subset of passes). thus if in my opinion while watching games if after receiving a pass a player had immediately shot and scored, i would count that as a "pass" for the player who threw it. passes thrown by players that in my opinion did not have any chance of being an assist i did not count as a "pass"....

so if on a particular possession a team had a half dozen passes in the backcourt without any defensive pressure whatsoever as they brought the ball up the floor to midcourt, since those passes in reality did not have any real opportunity to be an assist, i did not count those as "passes", i.e. they were superfluous. yet if for example a player was right underneath the basket with the ball, but passed it out away from the basket to a teammate at the top of the key who immediately shot the ball, that was counted as a "pass" because had the shot gone in the player throwing the pass would assuredly have been credited with an assist....

I know this is a critical part of Bob's simulator.

absolutely - if the computer doesn't how how often each player should handle the ball on offense, the stats won't come out right in re-creating what happened in real life. and the simulation does indeed re-create what happens in real life...

To me, it is just interesting. Very interesting. Bob's simulator and my individual win-loss records often end up similar through apparently very different methods.

The other thing I never sat down to think about is how his simulator partitions touches when players get blended together.

what's really neat about possession factor is just this - seeing how players change with new teammates. if you use the stats database look up for example adrian dantley. when dantley was traded from UTA to DET (8586/8687) his scoring dropped from 30 pts/g to less than 22 pts/g, playing about the same amount of minutes. but does anyone know why? it turns out that his touches/min simply decreased by about 25% (a large drop) as he went from a team with low touches/min players in two key positions (mark eaton and bob hansen) to a team with average touches/min at those two key positions (bill laimbeer and joe dumars). but - and this is key - what he did per touch, i.e. how often he shot, passed, got fouled, and turned the ball over - did not change (significantly) in going from one team to the other. now this in and of itself is not really a revelation, but at least it can be quantified with numbers. yes his stats per minute like FGA and FTA went down, but in reality he still did what he had always done - per touch - he simply had his touches reduced...
Back to top
View user's profile Send private message Send e-mail Visit poster's website
Dan Rosenbaum



Joined: 03 Jan 2005
Posts: 541
Location: Greensboro, North Carolina

PostPosted: Sat Feb 05, 2005 12:48 am Post subject: Re: Shouldn't possession usage count all potential assists? Reply with quote
Dan Rosenbaum wrote:
If we continued to give the passer one third of the credit for a potential assist, this would result in assists generating more than twice as much possession usage. This would result in point guards rising to the top of possession usage charts, which I think makes sense since they tend to control the ball much more than other players.
HoopStudies wrote:
I don't like to think about possessions this way, but touches, yes. PGs definitely touch the ball a ton, but the frequency with which they impact a final possession does tend to be small. Not universally small by any means. Magic Johnson was way up there. Marbury uses a lot. KJ did. Nash is up at 24% this year, behind Stoudemire.

I definitely think that twice as much "possession" usage would be wrong. 40% of a team's possessions being generated by a point guard just doesn't make sense with a lot of the offense being carried out through the other 4 guys.

I think you are exaggerating a bit here to make a point - although I worry that the exaggeration is serving to obscure the point. I am only giving one third of the usage to the potential assister (less after accounting for offensive rebounds), so getting to a 40 percent possession usage is nearly impossible for a pass-first point guard. Also, if at the end of the day, you feel like potential assisters are using too many possessions, all you have to do is adjust the credit downward from one third.
HoopStudies wrote:
But the point in the subject is correct -- it would be nice to have potential assists. The framework for individual possessions would be modified, including a term in the denominator and modifying the allocation of credit on it. (As I think I explained in the book, I had to assume about a 70-80% success rate on passes that would be assists. If we find something different, it can be modified.)

I am not sure where you made this assumption. As far as I can tell, the assist part in the numerator of your offensive rating divided by the assist part in the denominator is equal to 1 + 0.5 * (TMFG3M - FG3M) / (TMFGM - FGM). After multiplying by 100, this implies that every assist implicitly gets factored in with an offensive rating of between about 100 and 120. The higher ratings are for teams who make more three pointers. As far as I can tell, this is not very related to a team's true shooting percentage.

**We might also need to mulitply this by TMOR weight divided by TMOREB weight if these two weights are not the same. I had a hard time locating the definition of TMOR weight, so I was guessing that they were the same.

Dan Rosenbaum wrote:
Thus, a high percentage shooting PG would decrease his offensive rating by giving lots of assists to his low percentage shooting teammates. His possession usage would go up, but his offensive rating would go down.
HoopStudies wrote:
This is exactly what happens for Jason Kidd, for example. He is a poor shooting guard, which makes it all the more impressive that his teammates have shot reasonably well (not great).

This accounting was an important part of the method that I worked out in 1989, I remember. I was working a summer job at a national lab where the job didn't keep me all that busy, so I walked around with a basketball in my hand and worked on this theory. It makes the formulas much more complex, but it features the interaction that does happen in the game.

I apologize for not being able to figure out something that you figured out 16 years ago, but when I read through the formulas in BoP, the only difference I see between the assist parts in the numerator and denominator of your offensive rating is a function of the team three pointers made relative to field goals made. I don't see how playing for a poor shooting team affects Jason Kidd at all using your formula.
Back to top
View user's profile Send private message Send e-mail Visit poster's website Yahoo Messenger
Dan Rosenbaum



Joined: 03 Jan 2005
Posts: 541
Location: Greensboro, North Carolina

PostPosted: Sat Feb 05, 2005 1:04 am Post subject: Reply with quote
And Bob, now I see what touches are. They are the sum of field goal attempts, turnovers, free throws (multiplied by something like 0.44), and what I am calling potential assists. They are not really the same thing as DeanO's individual possessions, since you are potentially counting multiple touches in a single possession. You also are assuming that a turnover touch, a shooting touch, and a potential assist touch are equivalent, whereas that is not done in individual possession calculations.

One difficulty you must always face in your simulations is figuring out the relationship between touches per minute (or possession usage) and offensive efficiency. If a player like Adrian Dantley moves to a team where he touches the ball less, does his efficiency improve or does it stay about the same. DeanO in BoP suggest that most of the time there is an inverse relationship between touches (usage) and efficiency. But I would bet that relationship varies a lot from player to player and is very difficult to estimate from existing data.
Back to top
View user's profile Send private message Send e-mail Visit poster's website Yahoo Messenger
Kevin Pelton
Site Admin


Joined: 30 Dec 2004
Posts: 979
Location: Seattle

PostPosted: Sat Feb 05, 2005 2:08 am Post subject: Re: Shouldn't possession usage count all potential assists? Reply with quote
Dan Rosenbaum wrote:
Thus, a high percentage shooting PG would decrease his offensive rating by giving lots of assists to his low percentage shooting teammates.

I would argue that in some sense this is already happening because not only is this theoretical point guard (Steve Nash on the '04 Bobcats) not getting the assists he would be getting by passing to better players, he's also taking some associated risk of committing turnovers by throwing these passes that tends to reduce his rating.

But honestly, I'm searching for reasons not to like this because I'm still not ready to make the leap of faith that assists are not particularly valuable. It's going to take some time to wear down the traditional basketball notion of the importance of assists.

I already think that by counting assists equally in the numerator and the denominator of the offensive rating, we're giving assists too little credit.
Back to top
View user's profile Send private message Send e-mail Visit poster's website
HoopStudies



Joined: 30 Dec 2004
Posts: 706
Location: Near Philadelphia, PA

PostPosted: Sat Feb 05, 2005 3:22 am Post subject: Re: Shouldn't possession usage count all potential assists? Reply with quote
Dan Rosenbaum wrote:
Dan Rosenbaum wrote:
If we continued to give the passer one third of the credit for a potential assist, this would result in assists generating more than twice as much possession usage. This would result in point guards rising to the top of possession usage charts, which I think makes sense since they tend to control the ball much more than other players.
HoopStudies wrote:
I don't like to think about possessions this way, but touches, yes. PGs definitely touch the ball a ton, but the frequency with which they impact a final possession does tend to be small. Not universally small by any means. Magic Johnson was way up there. Marbury uses a lot. KJ did. Nash is up at 24% this year, behind Stoudemire.

I definitely think that twice as much "possession" usage would be wrong. 40% of a team's possessions being generated by a point guard just doesn't make sense with a lot of the offense being carried out through the other 4 guys.

I think you are exaggerating a bit here to make a point - although I worry that the exaggeration is serving to obscure the point.


Nope, not exaggerating. I just misunderstood your point. When you said you felt like assists would generate "twice the possession usage," I took it as individual possession usage. And I think you're saying that a factor of a third should go on top of that, so it would be an increase of about 6-7% for point guards on average. Is that right?

Dan Rosenbaum wrote:

HoopStudies wrote:
But the point in the subject is correct -- it would be nice to have potential assists. The framework for individual possessions would be modified, including a term in the denominator and modifying the allocation of credit on it. (As I think I explained in the book, I had to assume about a 70-80% success rate on passes that would be assists. If we find something different, it can be modified.)

I am not sure where you made this assumption. As far as I can tell, the assist part in the numerator of your offensive rating divided by the assist part in the denominator is equal to 1 + 0.5 * (TMFG3M - FG3M) / (TMFGM - FGM). After multiplying by 100, this implies that every assist implicitly gets factored in with an offensive rating of between about 100 and 120. The higher ratings are for teams who make more three pointers. As far as I can tell, this is not very related to a team's true shooting percentage.


The AST part on p 345 subracts off a player's own ability to shoot leaving you with how well the rest of the team shoots. It's not as good as using Roland's numbers, but I've actually plugged Roland's numbers in and typically not seen huge differences. Last year, for instance, when I use refined estimates for Steve Nash from Roland's data, his rating goes from 122 to 121. Marquis Daniels actually went up. (Where do you get that formula you have?)

Or are you saying something different?

Dan Rosenbaum wrote:
Thus, a high percentage shooting PG would decrease his offensive rating by giving lots of assists to his low percentage shooting teammates. His possession usage would go up, but his offensive rating would go down.


This is a concept we want, I agree. It penalizes poor decision makers. Having potential assists would help do this.

Dan Rosenbaum wrote:
When I read through the formulas in BoP, the only difference I see between the assist parts in the numerator and denominator of your offensive rating is a function of the team three pointers made relative to field goals made. I don't see how playing for a poor shooting team affects Jason Kidd at all using your formula.


I'm not sure we're understanding each other. The concept has been that Kidd's teammates shoot better when he's on the floor. The effective shooting percentage that determines the credit they get on the shots he assists on is determined by their shooting (including 3's), not Kidd's. So if Kidd is increasing the shooting percentage of his teammates, he gets relatively more credit for it. If they are shooting worse, he gets relatively less credit. That's the eq of p. 345. As I say, the estimate can be improved upon, but the essence is there. If we get potential assists, the credit for those can be apportioned as well.
_________________
Dean Oliver
Author, Basketball on Paper
The postings are my own & don't necess represent positions, strategies or opinions of employers.
Back to top
View user's profile Send private message Visit poster's website
Dan Rosenbaum



Joined: 03 Jan 2005
Posts: 541
Location: Greensboro, North Carolina

PostPosted: Sat Feb 05, 2005 9:32 am Post subject: Re: Shouldn't possession usage count all potential assists? Reply with quote
admin wrote:
Dan Rosenbaum wrote:
Thus, a high percentage shooting PG would decrease his offensive rating by giving lots of assists to his low percentage shooting teammates.

I would argue that in some sense this is already happening because not only is this theoretical point guard (Steve Nash on the '04 Bobcats) not getting the assists he would be getting by passing to better players, he's also taking some associated risk of committing turnovers by throwing these passes that tends to reduce his rating.

But honestly, I'm searching for reasons not to like this because I'm still not ready to make the leap of faith that assists are not particularly valuable. It's going to take some time to wear down the traditional basketball notion of the importance of assists.

I already think that by counting assists equally in the numerator and the denominator of the offensive rating, we're giving assists too little credit.

But I think that the true shooting percentage on potential assist shots is higher than the true shooting percentage on non-potential assist shots. If the difference is 10 percentage points (like I think Ed's study suggested), then by mulitiplying potential assists by two times the true shooting percentage for potential assists (in the numerator of the offensive rating), assists will tend to increase offensive ratings for all but really high-percentage shooters. Assists would be factored in with a rating of somewhere between 110 and 130.

Since, according to my reading, DeanO's formula actually factors in assists at a lower offensive rating and he counts assists less when computing possession usage, it appears to me that my formulation would give more, not less, credit to assisters.
Back to top
View user's profile Send private message Send e-mail Visit poster's website Yahoo Messenger
Dan Rosenbaum



Joined: 03 Jan 2005
Posts: 541
Location: Greensboro, North Carolina

PostPosted: Sat Feb 05, 2005 10:16 am Post subject: Re: Shouldn't possession usage count all potential assists? Reply with quote
HoopStudies wrote:
Nope, not exaggerating. I just misunderstood your point. When you said you felt like assists would generate "twice the possession usage," I took it as individual possession usage. And I think you're saying that a factor of a third should go on top of that, so it would be an increase of about 6-7% for point guards on average. Is that right?

I was misleading. The "twice the possession usage" is twice compared to a possession usage calculation that also uses the one third multiplier but does not account for potential assists. Since you, in essence, use a one half multiplier, my guess is that my formulation would only count assists a little more heavily than yours does. After factoring in the potential assists, my guess is that compared to your one half multiplier, my formulation uses a multiplier (1/3 * 1/potential assist true shooting percentage) something in the range between 0.53 and 0.61. This should lead to only a small difference from what you get - with the biggest differences coming for point guards on poor shooting teams.

Dan Rosenbaum wrote:
I am not sure where you made this assumption. As far as I can tell, the assist part in the numerator of your offensive rating divided by the assist part in the denominator is equal to 1 + 0.5 * (TMFG3M - FG3M) / (TMFGM - FGM). After multiplying by 100, this implies that every assist implicitly gets factored in with an offensive rating of between about 100 and 120. The higher ratings are for teams who make more three pointers. As far as I can tell, this is not very related to a team's true shooting percentage.
HoopStudies wrote:
The AST part on p 345 subracts off a player's own ability to shoot leaving you with how well the rest of the team shoots. It's not as good as using Roland's numbers, but I've actually plugged Roland's numbers in and typically not seen huge differences. Last year, for instance, when I use refined estimates for Steve Nash from Roland's data, his rating goes from 122 to 121. Marquis Daniels actually went up. (Where do you get that formula you have?)

Or are you saying something different?

Dan Rosenbaum wrote:
Thus, a high percentage shooting PG would decrease his offensive rating by giving lots of assists to his low percentage shooting teammates. His possession usage would go up, but his offensive rating would go down.

This is a concept we want, I agree. It penalizes poor decision makers. Having potential assists would help do this.

Dan Rosenbaum wrote:

When I read through the formulas in BoP, the only difference I see between the assist parts in the numerator and denominator of your offensive rating is a function of the team three pointers made relative to field goals made. I don't see how playing for a poor shooting team affects Jason Kidd at all using your formula.

I'm not sure we're understanding each other. The concept has been that Kidd's teammates shoot better when he's on the floor. The effective shooting percentage that determines the credit they get on the shots he assists on is determined by their shooting (including 3's), not Kidd's. So if Kidd is increasing the shooting percentage of his teammates, he gets relatively more credit for it. If they are shooting worse, he gets relatively less credit. That's the eq of p. 345. As I say, the estimate can be improved upon, but the essence is there. If we get potential assists, the credit for those can be apportioned as well.

According to your formulas on page 345 and 346 of BoP, the assist part of a scoring possession (after factoring in rebounding) can be written as:

scoring possession assist part (after factoring in rebounds) = X * Z,

where X = 0.5 * assist * [(TMPTS-TMFTM) - (PTS-FTM)]/[2*(TMFGA-FGA)]
Z = (1-TMOR/TMScPoss) * TMOR weight * TMPlay%

On page 348 of BoP, the assist part of points produced (after factoring in rebounding) can be writen as:

points produced assist part (after factoring in rebounds) = Y * X * Z,

where X & Z are as described above
Y = 2 * 0.5 * [(TMFGM-FGM) + 0.5 * (TMFG3M-FG3M)]/(TMFGM-FGM)

Y can be simplified to 1 + 0.5 * (TMFG3M-FG3M)/(TMFGM-FGM)

Since the only difference in the assist part in points produced and scoring possessions is Y, then Y is really the factor that determines how assists are factored into the offensive rating. It is mostly just a correction for how many three pointers a team makes.
Back to top
View user's profile Send private message Send e-mail Visit poster's website Yahoo Messenger
HoopStudies



Joined: 30 Dec 2004
Posts: 706
Location: Near Philadelphia, PA

PostPosted: Sat Feb 05, 2005 11:32 am Post subject: Re: Shouldn't possession usage count all potential assists? Reply with quote
Dan Rosenbaum wrote:
HoopStudies wrote:
Nope, not exaggerating. I just misunderstood your point. When you said you felt like assists would generate "twice the possession usage," I took it as individual possession usage. And I think you're saying that a factor of a third should go on top of that, so it would be an increase of about 6-7% for point guards on average. Is that right?

I was misleading. The "twice the possession usage" is twice compared to a possession usage calculation that also uses the one third multiplier but does not account for potential assists. Since you, in essence, use a one half multiplier, my guess is that my formulation would only count assists a little more heavily than yours does. After factoring in the potential assists, my guess is that compared to your one half multiplier, my formulation uses a multiplier (1/3 * 1/potential assist true shooting percentage) something in the range between 0.53 and 0.61. This should lead to only a small difference from what you get - with the biggest differences coming for point guards on poor shooting teams.



OK. That helps.

Dan Rosenbaum wrote:

According to your formulas on page 345 and 346 of BoP, the assist part of a scoring possession (after factoring in rebounding) can be written as:

scoring possession assist part (after factoring in rebounds) = X * Z,

where X = 0.5 * assist * [(TMPTS-TMFTM) - (PTS-FTM)]/[2*(TMFGA-FGA)]
Z = (1-TMOR/TMScPoss) * TMOR weight * TMPlay%

On page 348 of BoP, the assist part of points produced (after factoring in rebounding) can be writen as:

points produced assist part (after factoring in rebounds) = Y * X * Z,

where X & Z are as described above
Y = 2 * 0.5 * [(TMFGM-FGM) + 0.5 * (TMFG3M-FG3M)]/(TMFGM-FGM)

Y can be simplified to 1 + 0.5 * (TMFG3M-FG3M)/(TMFGM-FGM)

Since the only difference in the assist part in points produced and scoring possessions is Y, then Y is really the factor that determines how assists are factored into the offensive rating. It is mostly just a correction for how many three pointers a team makes.


Almost there, I think. So yes this is a correction to an assistant's teammates' field goal percentage by subtracting off his own ability to make shots. If a player is a better shooter, the ability of his teammates to shoot relative to him goes down and hence less weight on his assists. That sort of thing is useful to have. What you're suggesting and I'm not dismissing is adding in foul shots to this correction.
_________________
Dean Oliver
Author, Basketball on Paper
The postings are my own & don't necess represent positions, strategies or opinions of employers.
Back to top
View user's profile Send private message Visit poster's website

Ed Küpfer



Joined: 30 Dec 2004
Posts: 783
Location: Toronto

PostPosted: Sat Feb 05, 2005 11:41 am Post subject: Re: Shouldn't possession usage count all potential assists? Reply with quote
Dan Rosenbaum wrote:
I think we can estimate the number of potential assists.

potential assists = assists * (0.5 * points / field goals made) / true shooting percentage

Okay. I will repeat my experiment of charting potential and converted assists. I can probably do five games over the next couple of days. We'll see how well this holds up.

I'm curious as to why a player's shooting% would have anything to do with his non-assist passes. Or did you mean team shooting%? It's a long thread -- did I miss something?
_________________
ed
Back to top
View user's profile Send private message Send e-mail
Dan Rosenbaum



Joined: 03 Jan 2005
Posts: 541
Location: Greensboro, North Carolina

PostPosted: Sat Feb 05, 2005 1:34 pm Post subject: Re: Shouldn't possession usage count all potential assists? Reply with quote
Dan Rosenbaum wrote:

According to your formulas on page 345 and 346 of BoP, the assist part of a scoring possession (after factoring in rebounding) can be written as:

scoring possession assist part (after factoring in rebounds) = X * Z,

where X = 0.5 * assist * [(TMPTS-TMFTM) - (PTS-FTM)]/[2*(TMFGA-FGA)]
Z = (1-TMOR/TMScPoss) * TMOR weight * TMPlay%

On page 348 of BoP, the assist part of points produced (after factoring in rebounding) can be writen as:

points produced assist part (after factoring in rebounds) = Y * X * Z,

where X & Z are as described above
Y = 2 * 0.5 * [(TMFGM-FGM) + 0.5 * (TMFG3M-FG3M)]/(TMFGM-FGM)

Y can be simplified to 1 + 0.5 * (TMFG3M-FG3M)/(TMFGM-FGM)

Since the only difference in the assist part in points produced and scoring possessions is Y, then Y is really the factor that determines how assists are factored into the offensive rating. It is mostly just a correction for how many three pointers a team makes.
HoopStudies wrote:
Almost there, I think. So yes this is a correction to an assistant's teammates' field goal percentage by subtracting off his own ability to make shots. If a player is a better shooter, the ability of his teammates to shoot relative to him goes down and hence less weight on his assists. That sort of thing is useful to have. What you're suggesting and I'm not dismissing is adding in foul shots to this correction.

Perhaps I am misunderstanding your point, but Y is not a "correction to an assistant's teammates' field goal percentage by subtracting off his own ability to make shots." Y is a correction that accounts for teammates' made three pointers. I agree that your assist parts include corrections for an assistant's teammates' field goal percentage, but since that correction is in both the assist part in the numerator and the denominator of your offensive rating, it really doesn't have much impact on the offensive rating.

Since Y is the only thing that differs between your assist parts in the numerator and denominator of your offensive rating, it is the only thing, other than differences in possession usage due to assists, that determines how assists affect the offensive rating.

Thus, my formuation is more than just a foul shot correction. It is giving players who assist good shooting teammates a higher offensive rating. Ceteris paribus, comparing two players with the same number of assists on good shooting and bad shooting teams, the player on the good shooting team would have a higher offensive rating and lower possession usage. The player on the poor shooting team would have a lower offensive rating and higher possession usage. As far as I can tell, your formulation right now does not do that.
Back to top
View user's profile Send private message Send e-mail Visit poster's website Yahoo Messenger
Dan Rosenbaum



Joined: 03 Jan 2005
Posts: 541
Location: Greensboro, North Carolina

PostPosted: Sat Feb 05, 2005 1:55 pm Post subject: Re: Shouldn't possession usage count all potential assists? Reply with quote
Ed Küpfer wrote:
Dan Rosenbaum wrote:
I think we can estimate the number of potential assists.

potential assists = assists * (0.5 * points / field goals made) / true shooting percentage

Okay. I will repeat my experiment of charting potential and converted assists. I can probably do five games over the next couple of days. We'll see how well this holds up.

I'm curious as to why a player's shooting% would have anything to do with his non-assist passes. Or did you mean team shooting%? It's a long thread -- did I miss something?

I fixed the formula in the original post. It is teammates' true shooting percentage, preferably on potential assists.

Thanks for doing this charting. I think you can ignore scoring on non-shooting fouls, technicals, etc., since I think we can figure a way to take that out of the assist formulation.

It would probably be good to have a separate category for cases where there is practically no chance for an assist, such as breakaways or close-in offensive rebounds. We probably will need to include those in the non-potential assist category for these formulas, but it would be interesting to have that information for other discussions where we try to pin down the value of an assist.
Back to top
View user's profile Send private message Send e-mail Visit poster's website Yahoo Messenger

bchaikin



Joined: 27 Jan 2005
Posts: 676
Location: cleveland, ohio

PostPosted: Sun Feb 06, 2005 7:23 pm Post subject: Reply with quote
I would argue that in some sense this is already happening because not only is this theoretical point guard (Steve Nash on the '04 Bobcats) not getting the assists he would be getting by passing to better players, he's also taking some associated risk of committing turnovers by throwing these passes that tends to reduce his rating.

i'm not sure i quite understand this reference of steve nash on the bobcats, but if it refering to the assumption that if nash was on the bobcats he would be getting less assists, the assumpiton would be false. history shows a number players with high assist numbers or even high ast/min numbers on poor shooting teams...

nor do i quite understand the idea of "good" passers increasing the FG%s of the players they pass to (the assumption being they are "better" passes, i.e. they put the players receiving the passes in a better position to score) and how this relates to assists. the assumption may very well be true (i don't know), but i do know it doesn't correlate with assists...

i haven't been able to find a correlation between a player with a high number of assists and those assists leading to higher FG%s for the players receiving those passes. is there evidence of this at www.82games.com (if so this is just another example of why roland's data is so valuable)? having looked at the stats of a number of high assist PGs since 1977-78 being traded to new teams and looking at the FG% of that player's teammates before and after his arrival, i find no correlation for an increase in FG% being due to that high assist player...

right now steve nash and brevin knight are leading the league in assists per min with 0.32. no one else in the league is even close (jason williams and earl watson of the grizzlies are next best with 0.22 ast/min, along with knight's backup in charlotte jason hart). yet one PG - nash - plays for one of the best teams in the league and the other - knight - for one of the worst...

also right now the steve nash on court/off court eFG%s for the suns are .558 (1542 min) and .480 (770 min), and yes for jason kidd are .484 (1040 min) and .433 (1192 min), but for brevin knight are .448 (887 min) and .459 (1160 min). so while the data does show an increase in eFG% for "good" passers like nash and kidd, it does not for knight, who like nash is leading the league - by far - in ast/min...

one reason i believe steve nash is piling up assists at a career high rate in phoenix (and b.knight in charlotte) is the same reason andre miller did so in cleveland in 01-02. if a high touches/min PG who shoots less per touch than the average PG has starting teammates with average or below average touches/min for their positions (other than PG) but who also shoot the ball on a high 40%-50% of their touches or more will automatically get more assists, whether those teammates shoot a high FG% or not. why? because regardless of the FG%s they shoot they are still shooting alot once they receive that pass from the PG (i.e. they aren't often furthering their touch by dribbling, which usually takes away the PG's opportunity for an assist)...

once andre miller went to different teams (LAC/DEN) with frontcourt players or SGs with average or higher than average touches/min that could dribble and pass, his assists and ast/min numbers went down...

two of the best assist men in the league since 1977-78 were johnny moore and john lucas. yet both players had their highest ast/min marks on the same team - the 83-84 spurs. the reason is not because they were "better" passers that season, but it was because of the makeup of the team (the starters in particular). both mike mitchell and artis gilmore had lower touches/min than the average SF and C that season, yet each shot the ball with 50% or more of their touches. plus george gervin was taking 20 shots a game and shooting the ball with 50% of his touches...

plus the spurs that season were 6th in the league in FTM, which you would intuitively think would decrease assist opportunities. yet moore and lucas still had the highest ast/min marks in the league that season of all PGs...
Back to top
View user's profile Send private message Send e-mail Visit poster's website
Kevin Pelton
Site Admin


Joined: 30 Dec 2004
Posts: 975
Location: Seattle

PostPosted: Mon Feb 07, 2005 1:46 am Post subject: Reply with quote
What Dan is looking at provides us another way of looking at the assist success of players like Knight and Miller, in that, while they get plenty of assists per minute, their assist/assist opportunity ratio is almost certainly lower than Nash's. That's in part because their teammates' true shooting percentages are lower (dramatically so in Knight's case; not probably that huge of a difference for Miller), but it's a nice way of demonstrating the concept you're pointing out, that their assist rates are largely a product of the way their teams are put together.
Back to top
View user's profile Send private message Send e-mail Visit poster's website
Nikos



Joined: 16 Jan 2005
Posts: 346


PostPosted: Mon Feb 07, 2005 10:28 am Post subject: Reply with quote
admin wrote:
....... but it's a nice way of demonstrating the concept you're pointing out, that their assist rates are largely a product of the way their teams are put together.


Does that mean that assists do not mean as much from lets say the SG position?
Back to top
View user's profile Send private message
Kevin Pelton
Site Admin


Joined: 30 Dec 2004
Posts: 975
Location: Seattle

PostPosted: Mon Feb 07, 2005 10:44 am Post subject: Reply with quote
Another topic worth discussing is whether this kind of "assist success rate" is driven solely by teammates or by player skill in passing. Bob doesn't think so, but I disagree. A good passer will put his teammates in better situations to shoot. He'll deliver the ball to them where they want to catch it, and I do think that makes a difference.
Back to top
View user's profile Send private message Send e-mail Visit poster's website
HoopStudies



Joined: 30 Dec 2004
Posts: 705
Location: Near Philadelphia, PA

PostPosted: Mon Feb 07, 2005 2:47 pm Post subject: Re: Shouldn't possession usage count all potential assists? Reply with quote
Dan Rosenbaum wrote:
Thus, my formuation is more than just a foul shot correction. It is giving players who assist good shooting teammates a higher offensive rating. Ceteris paribus, comparing two players with the same number of assists on good shooting and bad shooting teams, the player on the good shooting team would have a higher offensive rating and lower possession usage. The player on the poor shooting team would have a lower offensive rating and higher possession usage. As far as I can tell, your formulation right now does not do that.


Actually my formulation does do that. There are offsetting effects that perhaps mute its impact, but the formula does do what you are saying it should. For instance, giving each of the other four starters an extra made shot in the Seattle-Charlotte game (changing efffg% from 58% to 64%) -- that on its own increases Luke's (he of 8 assists) offensive rating from 151 to 153. I checked this on season stats for a few teams, too, and saw the same thing.

The principal you mention is there in the formula, whether its strength is as large as one thinks it should be is something that can be handled through more theoretical development...
_________________
Dean Oliver
Author, Basketball on Paper
The postings are my own & don't necess represent positions, strategies or opinions of employers.
Back to top
View user's profile Send private message Visit poster's website
Dan Rosenbaum



Joined: 03 Jan 2005
Posts: 541
Location: Greensboro, North Carolina

PostPosted: Mon Feb 07, 2005 4:13 pm Post subject: Re: Shouldn't possession usage count all potential assists? Reply with quote
HoopStudies wrote:
Dan Rosenbaum wrote:
Thus, my formuation is more than just a foul shot correction. It is giving players who assist good shooting teammates a higher offensive rating. Ceteris paribus, comparing two players with the same number of assists on good shooting and bad shooting teams, the player on the good shooting team would have a higher offensive rating and lower possession usage. The player on the poor shooting team would have a lower offensive rating and higher possession usage. As far as I can tell, your formulation right now does not do that.

Actually my formulation does do that. There are offsetting effects that perhaps mute its impact, but the formula does do what you are saying it should. For instance, giving each of the other four starters an extra made shot in the Seattle-Charlotte game (changing efffg% from 58% to 64%) -- that on its own increases Luke's (he of 8 assists) offensive rating from 151 to 153. I checked this on season stats for a few teams, too, and saw the same thing.

The principal you mention is there in the formula, whether its strength is as large as one thinks it should be is something that can be handled through more theoretical development...

It might be useful to check if the offensive rating goes up for a player with zero assists when teammates' efg% rises. There are so many places in your offensive rating formula that teammates' efg% (or fg%) enters into the equation that this may have nothing to do with the assist part. It could increase the value of the one offensive rebound that Ridnour snared.

(I apologize for not doing this myself, but I have always worked with a simplified version of your possession usage and offensive ratings, so I cannot replicate your results exactly.)

Also, note that a six point increase in efg% is not far off from the range from the worst to best team in the NBA. And, according to your calculations, this only increases the offensive rating of a guy with 8 assists in 40 minutes from 151 to 153. For all practical purposes, this indicates teammate shooting has almost no effect on your offensive rating, even for high assisters.
Back to top
View user's profile Send private message Send e-mail Visit poster's website Yahoo Messenger
HoopStudies



Joined: 30 Dec 2004
Posts: 705
Location: Near Philadelphia, PA

PostPosted: Mon Feb 07, 2005 7:12 pm Post subject: Re: Shouldn't possession usage count all potential assists? Reply with quote
Dan Rosenbaum wrote:

It might be useful to check if the offensive rating goes up for a player with zero assists when teammates' efg% rises. There are so many places in your offensive rating formula that teammates' efg% (or fg%) enters into the equation that this may have nothing to do with the assist part. It could increase the value of the one offensive rebound that Ridnour snared.


It depends on the guy. Some guys had small upticks, others had no impact whatsoever. Other than the guys who were given the made field goal, though, Luke had the biggest increase in net points, getting essentially 0.6 points out of the 8 added to the team.

I'm not sure if this is the comparison we want to do, though. The concept I always had in mind is whether a guy helps his teammates actually increase their odds of scoring. In this case, it's through measured assists, but, as you say, it can be through getting foul shot opportunities, something I think is potentially important and I've avoided for a while. Adding theory to handle that is more important because...

Dan Rosenbaum wrote:
Also, note that a six point increase in efg% is not far off from the range from the worst to best team in the NBA. And, according to your calculations, this only increases the offensive rating of a guy with 8 assists in 40 minutes from 151 to 153. For all practical purposes, this indicates teammate shooting has almost no effect on your offensive rating, even for high assisters.


If the team's offensive rating went up under 8%, there is absolutely no indication whatsoever that Luke had anything to do with that 8% because his assist total didn't go up, and his offensive rating goes up about 1.5%, is that "too small"? I dunno.

If it is too small, it seems like you can argue for a while. I know MikeG actually didn't like my idea of giving more credit to good assistants, suggesting that it would just come out of good assistants having higher totals. We talked about this some years back and ultimately agreed to disagree.

But the discussion on putting foul shots in never did take place and I think is more fertile ground for theory improvements. If you want to sneak in something to change what I've done on team fg%, feel free. I'm open.
_________________
Dean Oliver
Author, Basketball on Paper
The postings are my own & don't necess represent positions, strategies or opinions of employers.
Back to top
View user's profile Send private message Visit poster's website
bchaikin



Joined: 27 Jan 2005
Posts: 676
Location: cleveland, ohio

PostPosted: Wed Feb 09, 2005 12:50 am Post subject: Reply with quote
Another topic worth discussing is whether this kind of "assist success rate" is driven solely by teammates or by player skill in passing. Bob doesn't think so, but I disagree. A good passer will put his teammates in better situations to shoot. He'll deliver the ball to them where they want to catch it, and I do think that makes a difference.

it's not that i don't think so, it's just that i don't see any way to show/prove this from assist totals. the assumption may very well be valid, but at one time (a number of years ago) i was asked to see if i could show it through the stats and i couldn't. actually i was asked if i could show that a good passer does increase the FG%s of players and from the data available i could not find proof of this....

i looked at PGs that went from team to team (via trades, free agency, etc) and looked at the FG%s of the players on the team they left (before and after the PG left) and the team they went to (before and after the PG got there) but could not find any definitive evidence or pattern that any PGs could increase players' FG%s...

having said that, the data from roland's website could give us a step forward on this. since he keeps track of the % of FGM by players that are assisted on, this can help. if over a few seasons of collecting this kind of data we find that there are certain players who consistently have high percentages of being assisted on their FGMs, and two or three of these kind of players end up on the same team (via trades, free agency, etc) and the PGs assists (or ast/min) go up, that can tell us its the players receiving the passes....

if on the other hand a PG is traded or signed and a number of players on that team have their percentage of assisted FGMs increase, that would tell us its the PG....
Back to top
View user's profile Send private message Send e-mail Visit poster's website
Dan Rosenbaum



Joined: 03 Jan 2005
Posts: 541
Location: Greensboro, North Carolina

PostPosted: Wed Feb 09, 2005 1:15 am Post subject: Re: Shouldn't possession usage count all potential assists? Reply with quote
HoopStudies wrote:
Dan Rosenbaum wrote:

It might be useful to check if the offensive rating goes up for a player with zero assists when teammates' efg% rises. There are so many places in your offensive rating formula that teammates' efg% (or fg%) enters into the equation that this may have nothing to do with the assist part. It could increase the value of the one offensive rebound that Ridnour snared.


It depends on the guy. Some guys had small upticks, others had no impact whatsoever. Other than the guys who were given the made field goal, though, Luke had the biggest increase in net points, getting essentially 0.6 points out of the 8 added to the team.

I still think this uptick for Ridnour was due to something other than Ridnour's assists. Among the non-starters, he was one of the leading scorers and I think may have been the leading offensive rebounder. According to your formula in BoP, the assist part of your offensive rating should have resulted in Ridnour's offensive rating going down (since the ratio of teammate made three pointers to made field goals fell). I think it was the other parts that offset that. If Ridnour had zero assists, would his offensive rating still have risen had his teammates had an extra made field goal? My guess is that there is a good chance that his offensive rating may have increased more than it did in your example.

I disagree with the claim that my formula is mostly just a free throw correction. It does do that, but I think that is a misleading interpretation of the difference between what you do and what I am suggesting. The assist part of your possession usage formula is a function of teammates' shooting, but your offensive rating is not, except through indirect, second-order effects through changes its effect on possession usage and through its effects on the teammates made three pointers relative to made field goal ratio. There are no direct effects of teammates' shooting that come through the assist part of your offensive rating formula - at least in the way it is described in BoP.
Back to top
View user's profile Send private message Send e-mail Visit poster's website Yahoo Messenger
Dan Rosenbaum



Joined: 03 Jan 2005
Posts: 541
Location: Greensboro, North Carolina

PostPosted: Wed Feb 09, 2005 1:23 am Post subject: Reply with quote
Bob, how come you completely ignore the charting data that Ed put together that suggested that true shooting percentages were about 10 percentage points higher on potentially assisted shots (shots that would have resulted in assists had the shot gone in) versus non-potentially assisted shots?

Now Ed is skeptical of that charting because of difficulties in charting assists and because of the possible non-representativeness of his Toronto game based sample, but still this is evidence that seems as good as about anything else that we could do. Comparisons of teams after trades is tricky because a team may have lost a scorer when adding an assister and the effects on teammates' shooting percentages may cancel each other out.

I don't think there is a lot of Roland's site to get at this question, because game logs do not measure potential assists. But someone more clever than I am may be able to come up with a way to get at this question with Roland's data.
Back to top
View user's profile Send private message Send e-mail Visit poster's website Yahoo Messenger
HoopStudies



Joined: 30 Dec 2004
Posts: 705
Location: Near Philadelphia, PA

PostPosted: Wed Feb 09, 2005 3:11 am Post subject: Re: Shouldn't possession usage count all potential assists? Reply with quote
Dan Rosenbaum wrote:

I disagree with the claim that my formula is mostly just a free throw correction. It does do that, but I think that is a misleading interpretation of the difference between what you do and what I am suggesting. The assist part of your possession usage formula is a function of teammates' shooting, but your offensive rating is not, except through indirect, second-order effects through changes its effect on possession usage and through its effects on the teammates made three pointers relative to made field goal ratio. There are no direct effects of teammates' shooting that come through the assist part of your offensive rating formula - at least in the way it is described in BoP.


And what I'm saying is feel free to improve upon it. I'm definitely willing to look at a proposed modification. Just probably not during the season. Sleep is already rare enough.
_________________
Dean Oliver

Mike G



Joined: 14 Jan 2005
Posts: 3491
Location: Hendersonville, NC

PostPosted: Tue Mar 11, 2008 9:51 am Post subject: White Boy Stats? Reply with quote
A friend visited me this past weekend. We'll call him Walter, since that's his name. I was showing Walter some of the stats I've compiled and created, since I've often mentioned them to him. He's one of the only long-term friends I've known who is a long-term NBA fan.

I showed him some stats at basketball-reference.com, and I explained that I don't really know what some of them mean. When he saw the list of top ORtg ('offensive rating') seasons (>1000 min.) -- Kerr '96, Hoiberg '05, Legler '96, Kerr '97 -- he just laughed.

"These are stats made by white boys, for white boys", he said. I was not able to disagree.

I don't know how many of us are 'white boys', or how relevant that is. I do see a curious over-representation of caucasian players in certain 'advanced stats'.

By rough count, only about 11% of the top 200 careers (in minutes) since 1977 have been white players, either American or Euro. I'd think any stat that places more than 11% of white players in the top 40, 100, whatever, suggests that white players somehow tend to excel in that area.

Among >1000-minute seasons, the fraction of whites among the top-20, 40, etc, for some 'advanced' stats:
Code:
seasons
top- ORtg DRtg Usg% Reb% Ast% TO% TMin
20 .45 .30 .00 .30 .70 .30 .05
40 .43 .28 .00 .23 .53 .25 .08
60 .43 .25 .00 .23 .42 .23 .15
80 .43 .24 .00 .25 .38 .20 .14
100 .39 .22 .00 .25 .32 .19 .11

John Stockton owns about half of the top 20 Ast% seasons. The highest Usg% season by a white player would be Paul Westphal, #104. The last column is 'total minutes', from which the 11% 'standard' was estimated.

If we look at whole careers, we see many more players among the top 100 in any category; no player can have more than one entry in a category. Among careers of at least 5000 minutes, the 'white' representation:

Code:
top- TMin ORtg DRtg Usg% Reb% Ast% TO%
20 .05 .45 .35 .00 .30 .20 .15
40 .08 .48 .33 .03 .25 .15 .23
60 .15 .45 .28 .07 .27 .15 .25
80 .14 .41 .28 .06 .25 .13 .23
100 .11 .38 .26 .09 .22 .14 .24
The top white (career) 'user' is Larry Bird, #38.

What explanations could account for the >.11 values in most of these categories? I realize many will not touch this discussion with a 10-foot slide rule, and it's cool. But on the chance that anyone should want to share introspection about this, it will be welcome.
_________________
`
36% of all statistics are wrong
Back to top
View user's profile Send private message Send e-mail
Ryoga Hibiki



Joined: 06 Oct 2007
Posts: 24


PostPosted: Tue Mar 11, 2008 10:29 am Post subject: Reply with quote
1) who are those white players? It doesn't look like a big sample too me
2) what positions do they play? I think that influences the rebounding numbers a lot

Btw, are we ready to accept that the top NBA athletes are black and that genetics play a big part in that?
that's where everything comes from, imo, because it writes the profile of what a white players needs to do to make it in the league.
Back to top
View user's profile Send private message
Harold Almonte



Joined: 04 Aug 2006
Posts: 616


PostPosted: Tue Mar 11, 2008 11:16 am Post subject: Reply with quote
Quote:
are we ready to accept that the top NBA athletes are black and that genetics play a big part in that?


That's something to be sorry, but it's real. Basketball game is very athletic, it's composed mostly of jumps and quick sprints and lateral moves, and the type of muscle fibers that help to do that depends on genetics. Where whites have, or could have, some advantage is in the eye-hand coordination part of the game, that is shooting and passing.
Back to top
View user's profile Send private message
Neil Paine



Joined: 13 Oct 2005
Posts: 774
Location: Atlanta, GA

PostPosted: Tue Mar 11, 2008 11:52 am Post subject: Re: White Boy Stats? Reply with quote
Mike G wrote:
When he saw the list of top ORtg ('offensive rating') seasons (>1000 min.) -- Kerr '96, Hoiberg '05, Legler '96, Kerr '97 -- he just laughed.

"These are stats made by white boys, for white boys", he said. I was not able to disagree.


I realize he was probably kidding, but this is just insulting. I know you don't know what ORtg means, Mike, you've made that clear on more than a few occasions, but for someone to suggest that it was created with racist motives is nothing short of idiocy.

A refresher: ORtg is the number of points produced by a player per 100 individual possessions used. It is a pure efficiency stat. It is not very useful to look at alone, without also considering some usage metric like the % of team possessions used while on the floor.

Take a gander (you and your friend) at the all-time best single-season 3-point shooting seasons. Go ahead, I'll wait. Look familiar? Damn, I guess this means 3-point shooting percentage is a stat made by white boys for white boys, too!Rolling Eyes

Or it could be the fact that ORtg is heavily skewed toward 3-point shooting specialists... When they shoot, they only have to maintain an eFG% of 33%, while 2-point shooters have to hit 50% to maintain the same level of efficiency. Plus, they rarely shoot non-3s, and they rarely shoot when they aren't open, which means their average points/possession is always sky-high. But looking at ORtg without any context (namely usage) is ignorant, because we've shown on multiple occasions that ORtg goes down when you take a bigger role in the offense. These "white boys" were using 12-15% of possessions while on the floor -- which was their role on the team -- while somebody else was doing the heavy lifting on offense. Everyone played their part for the betterment of the team. What's so hard to understand about that?

Now, I'll grant you the question of why a disproportionate number of 3-point shooting specialists are white. I don't have an answer there, and I'm not even going to try for one. But to suggest ORtg is racist because good 3-point specialists are the most efficient per-possession offensive players in the game is moronic.

Last edited by Neil Paine on Tue Mar 11, 2008 2:33 pm; edited 2 times in total
Back to top
View user's profile Send private message Visit poster's website
jkubatko



Joined: 05 Jan 2005
Posts: 702
Location: Columbus, OH

PostPosted: Tue Mar 11, 2008 11:55 am Post subject: Re: White Boy Stats? Reply with quote
Mike G wrote:
I showed him some stats at basketball-reference.com, and I explained that I don't really know what some of them mean. When he saw the list of top ORtg ('offensive rating') seasons (>1000 min.) -- Kerr '96, Hoiberg '05, Legler '96, Kerr '97 -- he just laughed.


Does Walter often laugh at his own ignorance?
_________________
Regards,
Justin Kubatko
Basketball-Reference.com
Back to top
View user's profile Send private message Send e-mail Visit poster's website
S.K.



Joined: 18 Feb 2005
Posts: 61
Location: Toronto

PostPosted: Tue Mar 11, 2008 11:56 am Post subject: Reply with quote
Ryoga Hibiki wrote:

Btw, are we ready to accept that the top NBA athletes are black and that genetics play a big part in that?
that's where everything comes from, imo, because it writes the profile of what a white players needs to do to make it in the league.


Genetics play a part, certainly, but the biggest X-factor that affects the racial makeup of any sport is CULTURE. Some athletes have skillsets which are best suited to one sport, but many exceptional athletes have several options available to them. If your friends play a certain sport, or your role models (ie father, big brother) are fans of a certain sport, you are much more likely to end up choosing that sport to focus on.

Looking specifically at North America, I think there's a very specific reason why the PF/C positions see a larger percentage of white players - if you're 6'11 or 7'0, there aren't many big-money sporting options available to you. It's pretty much basketball, or try to be a freak at hockey, football (or a pitcher in baseball, but that's a more specific skillset). If on the other hand you are 6'6 and have the skillset of an NBA swingman, you could also be an NHL defenseman, or a wide reciever/TE, or quarterback, or first baseman. Same argument if you're a 6'2 PG - you could be a dynamite center fielder or strong safety. Cultural factors MIGHT (I have no research to back this up, just going on my own observations) be more likely to push a black athlete to basketball, and say a white athlete to hockey. These are stereotypes, but I don't think it's controversial to say that most hockey players are white. Economic factors obviously come into play here as well - basketball is the cheapest sport to play, it requires only one ball per game and a net can be acquired or commandeered very easily.

So... and this is a very roundabout way of addressing Mike's point... I think there are legitimate reasons to posit that a white player in the NBA is more likely to succeed because of a SPECIFIC skill (shooting, dribbling, being tall) rather than just pure "athleticism". And these "advanced" stats like shooting efficiency or rebound rate tend to capture specific skills better than all-around athleticism.

That ended up being more rambling than I'd hoped. I hope no one is offended by anything in there - if anyone disagrees with my incredibly subjective reasoning I'd love a rebuttal.
_________________
No books - no articles - no website.
Just opinions.

Ill-informed opinions.
Back to top
View user's profile Send private message Send e-mail MSN Messenger
kjb



Joined: 03 Jan 2005
Posts: 860
Location: Washington, DC

PostPosted: Tue Mar 11, 2008 12:02 pm Post subject: Reply with quote
Just eyeballing the stats Mike posted, it would suggest that whites who played in the league during this era rebounded, passed and shot the ball reasonably well, but committed some turnovers and didn't shoot the ball very much. It would suggest that whites who made the league were likely either bigs (rebounds and turnovers), PGs (see assists and turnovers), or role-playing specialists (high ortg, low usage). The drtg numbers may also be related to the number of bigs and the rebounding.

Other than that, I largely agree with davis21wylie2121 and Justin's comments.
Back to top
View user's profile Send private message AIM Address Yahoo Messenger
Neil Paine



Joined: 13 Oct 2005
Posts: 774
Location: Atlanta, GA

PostPosted: Tue Mar 11, 2008 12:13 pm Post subject: Reply with quote
Exactly, most of these advanced stats try to reflect certain skills: Reb% for rebounding, Ast% for passing, etc. There are players out there who focus on nothing but 1 or 2 core skills -- we call them role players. Rebounding specialists, 3-point specialists, Shot-blocking specialists, Pure passing PGs, and so on and so forth. Perhaps a disproportionate number of whites are showing up in these categories because they had to focus on a specific role (at the expense of everything else) to survive in the league. Along those same lines, I would imagine very few of them show up in the Top 100 in more than 1 category, because being a specialist means you are not a good all-around player. In reality, Mike has just detected the presence of specialized role players, many of whom happen to be white.

So instead of looking at the advanced stats and asking why they are allegedly slanted toward whites, maybe we should step back and look at what they are measuring, and then ask why whites are overrepresented in certain "skill" categories (passing, rebounding, etc.), and what that tells us about the racial makeup of the league.
Back to top
View user's profile Send private message Visit poster's website
Charles



Joined: 16 May 2005
Posts: 134


PostPosted: Tue Mar 11, 2008 12:18 pm Post subject: Reply with quote
Interesting observations Mike. It's certainly a legitimate inquiry. The problem is discerning causal relationships in this kind of social data. Do you think race and playing style could both be driven by more fundamental social, cultural and demographic factors? I suspect they are.

Perhaps athletic individuals raised in lower economic environments (regardless of race) are more likely to see sports as a viable route to college or even a profession. Therefore, "usage" type stats - which may be seen as more likely to garner scholarship offers might logically, in some cases, be emphasized. As far as "genetic" factors go. I don't buy that at all. Genetics is invoked far too loosely and with far too little evidence in these things.
Back to top
View user's profile Send private message
JMartin437



Joined: 05 Mar 2008
Posts: 1
Location: New York City

PostPosted: Tue Mar 11, 2008 12:20 pm Post subject: Re: White Boy Stats? Reply with quote
Mike G wrote:
A friend visited me this past weekend. We'll call him Walter, since that's his name. I was showing Walter some of the stats I've compiled and created, since I've often mentioned them to him. He's one of the only long-term friends I've known who is a long-term NBA fan.


"These are stats made by white boys, for white boys", he said. I was not able to disagree.


Just for the record, I peruse this forum daily, use advanced stats in my own writing for the NY Sun, and on the basis of my facility with them I was recently hired by a black news and cultural affairs site, and I am most assuredly not white.
I AM male, so Walter may have been half right in his ignorant characterization.

-MJ
Back to top
View user's profile Send private message Send e-mail
Mike G



Joined: 14 Jan 2005
Posts: 3491
Location: Hendersonville, NC

PostPosted: Tue Mar 11, 2008 12:53 pm Post subject: Reply with quote
While I was out, it occurred to me that there might actually be an inverse relationship between Usg% and ORtg. At least, from a sample of, say, players with >10,000 career minutes. Anyone know how to check this?

In general, it seems that teams would have their most efficient scorers using the most possessions. There's a theory that certain low-%, hi-usage players can enhance teammates' efficiencies; and yet other hi-use/hi-% players may cut into team efficiency.

I do think everyone replying in this thread has said and/or asked something of value. The typical white NBA player is probably bigger and/or slower than the typical black player. Jumping we know about.

Certain ratios are more natural: FT/FTA, for example. (Blk+Ast)/FTA would be less natural, and a reasonable person could wonder how it could be useful. There's no limit to the combinations of such ratios. Pts/(2*FGA+.88*FTA) is a useful ratio, according to many.[/list]
_________________
`
36% of all statistics are wrong
Back to top
View user's profile Send private message Send e-mail
Ryoga Hibiki



Joined: 06 Oct 2007
Posts: 24


PostPosted: Tue Mar 11, 2008 12:53 pm Post subject: Reply with quote
S.K. wrote:
Genetics play a part, certainly, but the biggest X-factor that affects the racial makeup of any sport is CULTURE. Some athletes have skillsets which are best suited to one sport, but many exceptional athletes have several options available to them. If your friends play a certain sport, or your role models (ie father, big brother) are fans of a certain sport, you are much more likely to end up choosing that sport to focus on.

I'll tell you my problem in disregarding genetic reasons: I can accept the answer that 88% of NBA are blacks because of role models, less opportunities etc.
What's wrong is how those groups are shaped. Can we agree that in the top athletes as far as eaping abilities and explosiveness there are more blacks than whites in a proportion way different than 88%?
I see those as mostly god given abilities, so the only explanation I see is that certain groups are more likely to develop them.
I hope we see the difference between saying that black players are better, and saying that they're consistently better as far as pure athletic abilities. The names of Nowitzki or Nash don't change things a bit, they're there despite the fact they don't outjump or overpower anyone.
Quote:

So... and this is a very roundabout way of addressing Mike's point... I think there are legitimate reasons to posit that a white player in the NBA is more likely to succeed because of a SPECIFIC skill (shooting, dribbling, being tall) rather than just pure "athleticism". And these "advanced" stats like shooting efficiency or rebound rate tend to capture specific skills better than all-around athleticism.

at the ned we reach the same conclusion, even we see different reasons behind it: a white player to make it into the NBA is often a shooting specialist ho needs to be really efficient to play.

I hope nothing sounds offensive, I'm not used to deal with racial discussions as in my country there's no real integration yet.
Back to top
View user's profile Send private message
Ryoga Hibiki



Joined: 06 Oct 2007
Posts: 24


PostPosted: Tue Mar 11, 2008 1:06 pm Post subject: Reply with quote
I found this, the original article is nomore available:

Dawkins on hoops in black and white
Charley Rosen / Special to FOXSports.com

Darryl Dawkins has never worried about being politically correct. Ever since he made headlines in 1975 when he became the first schoolboy to be drafted into the NBA (Philadelphia — fifth pick overall), Dawkins was never shy about speaking his mind.

According to Dawkins, the outcome of the upcoming Olympic basketball tournament hinges on a subject no one else dares to fully address — the racial components that define basketball as we know it.

"The game is the same," says Dawkins. "The object is for the good guys to score and to keep the bad guys from scoring. But there's a big difference between black basketball and white basketball."

Growing up poor (but happy) near Orlando, Fla., Dawkins learned the former before he learned the latter. "Black basketball is much more individualistic," he says. "With so many other opportunities closed to young black kids, the basketball court in the playground or the schoolyard is one of the few places where they can assert themselves in a positive way. So if somebody makes you look bad with a shake-and-bake move, then you've got to come right back at him with something better, something more stylish. And if someone fouls you hard, you've got to foul him even harder. It's all about honor, pride, and establishing yourself as a man."

Once the black game moves indoors and becomes more organized, the pressure to establish bona fides increases. "Now you're talking about high school hoops," says Dawkins. "So if you're not scoring beaucoup points, if your picture isn't in the papers, if you don't have a trophy, then you ain't the man and you ain't nothing. Being second-best is just as bad as being last. And if a teammate hits nine shots in a row, the black attitude is, 'Screw him. Now it's my turn to get it on.'"

If young black players usually cherish untrammeled creativity, white hooplings mostly value more team-oriented concepts. "White basketball means passing the hell out of the ball," says Dawkins. "White guys are more willing to do something when somebody else has the ball — setting picks, boxing out, cutting just to clear a space for a teammate, making the pass that leads to an assist pass. In white basketball, there's a more of a sense of discipline, of running set plays and only taking wide open shots. If a guy gets hot, he'll get the ball until he cools off."

Why is white basketball so structured and team-oriented?

"Because the white culture places more of a premium on winning," Dawkins believes, "and less on self-indulgent preening and chest-beating. That's because there are so many other situations in the white culture where a young kid can express himself."

As the twig is bent, so grows the tree. When Dawkins and the Sixers squared off against the Portland Trail Blazers for the NBA championship in 1977, Philadelphia's most dynamic players were Julius Erving, George McGinnis, World B. Free, and Dawkins.

"They beat us in six games," Dawkins recalls, "and the series marked the most blatant example of the racial difference in NBA game plans. We were much more flamboyant than Portland, and certainly more talented. We had more individual moves, more off-balance shots, more fancy passes, more dunks, and more entertaining stuff. But everybody wanted to shoot and be a star (including me), and nobody was willing to do the behind-the-scenes dirty work."

Meanwhile, the white players at the core of Portland's eventual success were Dave Twardzik, Bobby Gross, Larry Steele and Bill Walton. Dawkins notes that "Even the black guys like Lionel Hollins, Mo Lucas, Johnny Davis, Lloyd Neal played disciplined, unselfish white basketball. Credit their coach, Jack Ramsay, for getting everybody on the same page."

As much as Dawkins respected Portland's game plan, however, he was never crazy about Walton. "The guy was a good player who could really pass and had a nice jump hook," Dawkins opines. "What made Walton so effective was that he was surrounded by talented players who wanted to win and weren't concerned with being stars. Personally, I think that Walton was, and still is, full of baloney. Back then, he had this mountain-man image, he smoked lots of pot, and I don't think he bathed regularly. And the league let him play with a red bandana tied around his head. To say nothing of his involvement with Patty Hearst.

"If a black player ever tried any of that kind of stuff he would've been banished from the NBA in a heartbeat. Yet in spite of all the messed up things Walton did as a player, now that he's a TV announcer all he does is tear down everybody else. The guy still ticks me off."

During his 15-year tenure in the NBA, Dawkins' signature move was bulldozing to the basket and smashing the Plexiglas backboard to smithereens. He was brash, outlandish, funny, and irresistible. He called himself "Chocolate Thunder," claimed to be from the planet Lovetron, and devised names for his more awesome dunks — among the most noteworthy were In Your Face Disgrace, Cover Yo Damn Head, Sexophonic Turbo Delight, and his classic If You Ain't Groovin' Best Get Movin'-Chocolate Thunder Flyin'-Robinzine Cryin'-Teeth Shakin'-Glass Breakin'-Rump Roastin'-Bun Toastin'-Glass Still Flyin' Wham-Bam-I-Am Jam!

For the past four years, the 6-foot-11, 285-pound Dawkins has been coaching the Pennsylvania Valley Dawgs in the summertime United States Basketball League. In so doing, he's won two championships (2002 and 2004) and distinguished himself as a superior motivator and big man coach, as well as the kind of on- and off-court teacher who can help transform wild young hooplings into mature gamers. As a by-product of his own maturation, Dawkins can also see the pluses and minuses of both black and white basketball.

"The black game by itself," he says, "is too chaotic and much too selfish. No one player is good enough to beat five opponents on a consistent basis. The black style also creates animosities among the players because everybody ends up arguing about who's shooting too much and who's not shooting enough."

But the white game also has its drawbacks: "It can get too predictable and even too cautious because guys can be afraid to take risks and make mistakes."

Dawkins believes that the best NBA teams combine the best of both. "In basketball and in civilian life," Dawkins says, "freedom without structure winds up being chaotic and destructive. Only when it operates within a system can freedom create something worthwhile."

And, according to Dawkins, this is the most difficult task at hand for Larry Brown. "Only Tim Duncan and Carlos Boozer are willing to play white basketball. All the other guys on Team USA really want to go off on their own.

"Unless Brown can bleach some of the selfish funk from their game, they'll be lucky to win the bronze."

Charley Rosen, former CBA coach, author of 12 books about hoops, the next one being A PIVOTAL SEASON -- HOW THE 1971-72 LA LAKERS CHANGED THE NBA, is a frequent contributor to FOXSports.com.
Back to top
View user's profile Send private message
Chicago76



Joined: 06 Nov 2005
Posts: 98


PostPosted: Tue Mar 11, 2008 1:12 pm Post subject: Reply with quote
I won't touch anything relating to biology on here, but I will bring up a couple of social issues. I think we can all agree that society is segregated at least by economics/class and generally by race. I'm a soccer fan as well, and you often hear about different styles of play depending upon culture/country.

England = fast, direct, and physical
Germany = disciplined tactics, more physical players
Italy = rock solid defense, counter attacks
Brazil = flair, do everything players
Dutch = closest thing to Brazilian style in Europe

It's not to say that a Dutchman might not play like an Italian or a German, but in general, there are unique styles.

In a sense we could just be seeing that white players and black players within the US learn their games in two different countries within a country. Sometimes you get interesting, mixed results, like an Eric Gordon. He grew up in a predominantly white part of suburban Indianapolis in an integrated high school. While he is definitely known to take the ball to the basket, he is generally a shooter at Indiana w/ low rebounding and assist totals.
Back to top
View user's profile Send private message
S.K.



Joined: 18 Feb 2005
Posts: 61
Location: Toronto

PostPosted: Tue Mar 11, 2008 1:45 pm Post subject: Reply with quote
Chicago76 wrote:
I won't touch anything relating to biology on here, but I will bring up a couple of social issues. I think we can all agree that society is segregated at least by economics/class and generally by race. I'm a soccer fan as well, and you often hear about different styles of play depending upon culture/country.

England = fast, direct, and physical
Germany = disciplined tactics, more physical players
Italy = rock solid defense, counter attacks
Brazil = flair, do everything players
Dutch = closest thing to Brazilian style in Europe

It's not to say that a Dutchman might not play like an Italian or a German, but in general, there are unique styles.


The soccer analogy is interesting - this reminded me (as a Canadian) of the international integration of the NHL in the 70s, 80s, and 90s. There were very specific expectations of a player based on what country he was from, and even now your typical Russian player is expected to be more creative and less physical than your typical Canadian player. Individual players obviously vary but until fairly recently international competitions featured national teams with very distinct styles.
_________________
No books - no articles - no website.
Just opinions.

Ill-informed opinions.

Charles



Joined: 16 May 2005
Posts: 134


PostPosted: Tue Mar 11, 2008 1:51 pm Post subject: Reply with quote
Ryoga Hibiki wrote:

I'll tell you my problem in disregarding genetic reasons: I can accept the answer that 88% of NBA are blacks because of role models, less opportunities etc.
What's wrong is how those groups are shaped. Can we agree that in the top athletes as far as eaping abilities and explosiveness there are more blacks than whites in a proportion way different than 88%?

Ryoga, I have no way of knowing whether your assumption about leaping ability is correct. However, NBA basketball players and sub-groups of players are, to an extent, self-selected. And it is certainly likely that - as Dawkins points out (as only he can) - they are products of different development environments. So, even if your assumption were true it would not tell us anything about genetic pools. You just can't drag genetics from the lab to the street.

Incidentally, I think elite volleyball players have measured verticals as high or higher than elite basketball players. Yet Olympic volleyball is very competitive among U.S., South American, Eastern European, and Asian countries. I believe Steve Timmons had a higher [standing] vertical than Vince Carter.

Thanks for the quote from Chocolate Thunder. Very interesting and one of my all time favorite ballers.
Back to top
View user's profile Send private message
asimpkins



Joined: 30 Apr 2006
Posts: 245
Location: Pleasanton, CA

PostPosted: Tue Mar 11, 2008 2:18 pm Post subject: Reply with quote
S.K. wrote:
Genetics play a part, certainly, but the biggest X-factor that affects the racial makeup of any sport is CULTURE.


It's anecdotal, but I remember thinking about this while watching Through The Fire, the documentary on Sebastian Telfair. Not just his family and friends, but the entire community was fired up and supportive of this single-minded, obsessive goal to make it to the NBA. It was incredible to see.

My experience couldn't have been more different. If I decided I wanted to play in the NBA and I was going to do nothing but develop basketball skills I would have had my entire world of adult role models telling me that basketball success was unlikely I that I needed to balance my life with more realistic career plans and an almost effortless path to college. Sure I could still try to play basketball, and they'd encourage me, but there would be a Plan B and Plan C and so on. It would be difficult to reach the same ultimate level of commitment.

Maybe the differences between our environments aren't as stark as I presented here -- or learned from watching a movie. But it seems to me very possible that the values of a community could play a huge role in the quantity and quality of players that they produce.
Back to top
View user's profile Send private message
Mike G



Joined: 14 Jan 2005
Posts: 3597
Location: Hendersonville, NC

PostPosted: Tue Mar 11, 2008 2:48 pm Post subject: Reply with quote
OK, let's look at Fred Hoiberg's 2005. His ORtg was 138, for a Min team that had an ORtg of 108. Next-best was KG's 117. So Hoiberg was (30/9) 3.3 times as good an option as Garnett, relative to the team's rate. According to this 'traditional' offensive-rating.

Fred's Usg% was 12.7% -- average is 20% -- less than half that of Garnett's and Cassell's; lower than Trent Hassell's. To me, that says he effectively 'did nothing' on 7.3% of possessions. Not his possessions, but team possessions.

If a player is not participating, effectively, 2/3 as often as he 'uses' a possession, then he's only a factor on 60% (3/5) of team possessions. Then we could suppose .60*138 = 83 was Hoiberg's 'real' ORtg.

Or we could split the difference: (138+83)/2 = 110.5

This is just 2.3 above the team avg. His 106 DRtg is 0.6 better than the team avg. Perhaps this is more in line with the minutes he received, from 2 coaches that year (Saunders and McHale).
_________________
`
36% of all statistics are wrong
Back to top
View user's profile Send private message Send e-mail
jkubatko



Joined: 05 Jan 2005
Posts: 702
Location: Columbus, OH

PostPosted: Tue Mar 11, 2008 3:02 pm Post subject: Reply with quote
Mike G wrote:
OK, let's look at Fred Hoiberg's 2005. His ORtg was 138, for a Min team that had an ORtg of 108. Next-best was KG's 117. So Hoiberg was (30/9) 3.3 times as good an option as Garnett, relative to the team's rate. According to this 'traditional' offensive-rating.

Fred's Usg% was 12.7% -- average is 20% -- less than half that of Garnett's and Cassell's; lower than Trent Hassell's. To me, that says he effectively 'did nothing' on 7.3% of possessions. Not his possessions, but team possessions.

If a player is not participating, effectively, 2/3 as often as he 'uses' a possession, then he's only a factor on 60% (3/5) of team possessions. Then we could suppose .60*138 = 83 was Hoiberg's 'real' ORtg.

Or we could split the difference: (138+83)/2 = 110.5

This is just 2.3 above the team avg. His 106 DRtg is 0.6 better than the team avg. Perhaps this is more in line with the minutes he received, from 2 coaches that year (Saunders and McHale).


Now we know why Walter was confused...
_________________
Regards,
Justin Kubatko
Basketball-Reference.com
Back to top
View user's profile Send private message Send e-mail Visit poster's website
Mike G



Joined: 14 Jan 2005
Posts: 3597
Location: Hendersonville, NC

PostPosted: Tue Mar 11, 2008 4:40 pm Post subject: Reply with quote
Well, confusion is cheap, and it's quick. Improvement takes some time and effort. If 'unused' possessions are part of the equation, then ORtg (Pts/Poss-used) may have some first-impression impact other than 'you gotta be kidding me'.

My first observation of a basketball fan's first take on 'advanced stats' was pretty enlightening. I could hope for better, but not really expect such. Walter has a PhD and has moved on (New Guinea), likely without any further thought to our petty machinations.

Idiocy, ignorance, moronic do not apply here. Relevance to reality may or may not.
_________________
`
36% of all statistics are wrong
Back to top
View user's profile Send private message Send e-mail
Mike G



Joined: 14 Jan 2005
Posts: 3597
Location: Hendersonville, NC

PostPosted: Wed Mar 12, 2008 5:17 am Post subject: Re: White Boy Stats? Reply with quote
davis21wylie2121 wrote:
...I guess this means 3-point shooting percentage is a stat made by white boys for white boys, too!
.

When the 3-pt distance was introduced, then shortened, it was (at least in part) justified by the league's desire to make less-athletic players more important. Many guys were making great things happen in college ball, and weren't able to contribute at the NBA level.

So yes, the 3-pt shot exists in part, or mostly, as an accomodation for the less quick/strong/athletic players out there.

So why do we call a stat 'offensive rating' when it doesn't rate a player's offensive ability? That's the obvious question, not answered in the stat pages.
Why not account for the player's 'usage', rather than list that as a separate stat?
_________________
`
36% of all statistics are wrong
Back to top
View user's profile Send private message Send e-mail
jkubatko



Joined: 05 Jan 2005
Posts: 702
Location: Columbus, OH

PostPosted: Wed Mar 12, 2008 7:05 am Post subject: Re: White Boy Stats? Reply with quote
Mike G wrote:
So why do we call a stat 'offensive rating' when it doesn't rate a player's offensive ability? That's the obvious question, not answered in the stat pages.
Why not account for the player's 'usage', rather than list that as a separate stat?


You seem to bring this up about once every six months. People have tried to explain this to you in the past, but you just don't make the effort to understand. I think it's time to move on to something else.
_________________
Regards,
Justin Kubatko
Basketball-Reference.com
Back to top
View user's profile Send private message Send e-mail Visit poster's website
Harold Almonte



Joined: 04 Aug 2006
Posts: 616


PostPosted: Wed Mar 12, 2008 9:34 am Post subject: Reply with quote
http://www.jonentine.com/reviews/AOL_Wh ... atters.htm
Back to top
View user's profile Send private message
Statman



Joined: 20 Feb 2005
Posts: 242
Location: Arlington, Texas

PostPosted: Wed Mar 12, 2008 2:51 pm Post subject: Reply with quote
Mike G wrote:
OK, let's look at Fred Hoiberg's 2005. His ORtg was 138, for a Min team that had an ORtg of 108. Next-best was KG's 117. So Hoiberg was (30/9) 3.3 times as good an option as Garnett, relative to the team's rate. According to this 'traditional' offensive-rating.

Fred's Usg% was 12.7% -- average is 20% -- less than half that of Garnett's and Cassell's; lower than Trent Hassell's. To me, that says he effectively 'did nothing' on 7.3% of possessions. Not his possessions, but team possessions.

If a player is not participating, effectively, 2/3 as often as he 'uses' a possession, then he's only a factor on 60% (3/5) of team possessions. Then we could suppose .60*138 = 83 was Hoiberg's 'real' ORtg.

Or we could split the difference: (138+83)/2 = 110.5

This is just 2.3 above the team avg. His 106 DRtg is 0.6 better than the team avg. Perhaps this is more in line with the minutes he received, from 2 coaches that year (Saunders and McHale).


Quick thought

I personally would take his rating times 12.7/20, and add the team rating minus his rating times 7.3/20 to get his new offensive rating tied to usage.

Or MAYBE you can just find out the difference in team ORating with the player and without him - and extrapolate a new rating if he played exactly 20% of the team minutes. Kevin Garnett would obviously end up with a higher offensive score than Hoiberg - with the larger amount of total minutes and greater usage with ORat higher than team average.

Of course - the problem with the above idea is that high usage players that end up below team average Orat will look even worse in relation to low usage players with a below team ORat.

Eh, whatever - I just woke up.
_________________
Dan

My current national college player rankings (and other stuff):
http://www.pointguardu.com/f136/statman ... post355594
Back to top
View user's profile Send private message Send e-mail
Harold Almonte



Joined: 04 Aug 2006
Posts: 616


PostPosted: Thu Mar 13, 2008 10:56 am Post subject: Reply with quote
Quote:
Mike G wrote:
So why do we call a stat 'offensive rating' when it doesn't rate a player's offensive ability? That's the obvious question, not answered in the stat pages.
Why not account for the player's 'usage', rather than list that as a separate stat?


You seem to bring this up about once every six months. People have tried to explain this to you in the past, but you just don't make the effort to understand. I think it's time to move on to something else.


We always want that a measure of performance give us an accurate look of the skill or ability level at the same time, and yes skill=efficiency*usage, but we're still discussing about the each skill's tradeoff, and each individual's tradeoff, and performance is the sum of multiple skills. It's never unworthy to discuss, and rediscuss, and rediscuss, the value of assists, ballhandling help, passing help, and not passing help against scoring, inside the offensive squeme and offensive rating, to remember us who are the offensive leaders between the Nash-es and the Stoudemire-s, and between the Calderon-s and the Bosh-s.
Back to top
View user's profile Send private message
Mike G



Joined: 14 Jan 2005
Posts: 3597
Location: Hendersonville, NC

PostPosted: Thu Mar 13, 2008 1:24 pm Post subject: Reply with quote
Statman wrote:

I personally would take his rating times 12.7/20, and add the team rating minus his rating times 7.3/20 to get his new offensive rating tied to usage.
...

Without parentheses, I can't follow this formula.

Usg% is clearly % of possessions used; ORtg is efficiency of possessions used. First, I'm going to call that latter quantity 'possession efficiency' (PEf). Then we can look for a 'new offensive rating' incorporating both measures.

As a first pass, I propose:
ORtg (new) = PEf*Usg%^N

I've divided PEf by the league average (106.9), and Usg% by 20, to come up with this version of ORtg. Value of 1.22 is 122% of league average.

Of the top 100 players in minutes this year, here are the top 20 ORtg (new) when N = .10, .30, and .50 , for examples.
Code:
ORtg N=.10 PEf Usg% ORtg N=.30 PEf Usg% ORtg N=.50 PEf Usg%
1.22 Billups 1 43 1.28 James 15 1 1.43 James 15 1
1.19 Paul 3 21 1.26 Stoudemire 5 11 1.35 Stoudemire 5 11
1.18 Stoudemire 5 11 1.26 Billups 1 43 1.34 Bryant 28 3
1.18 Calderon 2 79 1.25 Paul 3 21 1.32 Nowitzki 10 7
1.16 James 15 1 1.23 Nowitzki 10 7 1.32 Ginobili 19 5

1.15 Nowitzki 10 7 1.22 Bryant 28 3 1.32 Paul 3 21
1.15 Bosh 11 10 1.22 Bosh 11 10 1.31 Bosh 11 10
1.13 RBrewer 4 82 1.22 Ginobili 19 5 1.29 Billups 1 43
1.13 Nash 7 47 1.17 Redd 33 9 1.25 Redd 33 9
1.13 Ginobili 19 5 1.17 Iverson 30 13 1.25 Iverson 30 13

1.12 Bryant 28 3 1.16 Garnett 17 33 1.24 Anthony 71 4
1.11 Gasol 13 48 1.16 Nash 7 47 1.24 Wade 89 2
1.11 Garnett 17 33 1.15 Maggette 37 17 1.23 Maggette 37 17
1.11 Ellis 12 54 1.15 Calderon 2 79 1.22 Duncan 54 8
1.11 Stojakovic 9 66 1.15 Ming 42 14 1.22 Ming 42 14

1.11 DLee 6 91 1.14 DHoward 24 35 1.21 Garnett 17 33
1.10 DWilliams 18 42 1.14 Anthony 71 4 1.21 Boozer 44 15
1.10 Iverson 30 13 1.14 DWilliams 18 42 1.20 AJefferson 57 12
1.10 DHoward 24 35 1.14 Gasol 13 48 1.19 RJefferson 51 18
1.10 Redd 33 9 1.14 Boozer 44 15 1.19 Roy 36 30

I've listed players' rank (1-100) in pre-existing categories. Note the low exponent (.10, the first group) is dominated by high-PEf guys; the higher exponent favors the high-Usg players.
_________________
`
36% of all statistics are wrong

Last edited by Mike G on Fri Mar 14, 2008 12:22 pm; edited 2 times in total
Back to top
View user's profile Send private message Send e-mail
Neil Paine



Joined: 13 Oct 2005
Posts: 774
Location: Atlanta, GA

PostPosted: Thu Mar 13, 2008 2:00 pm Post subject: Reply with quote
Mike, why don't you just sit down and (re?) read Basketball on Paper? The whole point of ORtg and %Poss is that they are separated out into 2 numbers... Together, they tell us how big a player's role is in the offense, and how efficient they have been playing that role. There's no need to create a "new" ORtg by raising ORtg to some arbitrary exponent. If you really wanted to "adjust" ORtg, you could estimate what a player's ORtg would be at different levels of usage by using the traditional tradeoffs:

Code:
Initial %Poss Sensitivity
---------------------------
<18% +/- 0.8
18-23% +/- 0.6
>23% +/- 0.4


Or even Eli's new tradeoff of +/-1.25.

But I would recommend sitting down and reading Basketball on Paper first, because it looks like you don't understand the purpose and/or meaning of these stats right now.
Back to top
View user's profile Send private message Visit poster's website
Mike G



Joined: 14 Jan 2005
Posts: 3597
Location: Hendersonville, NC

PostPosted: Thu Mar 13, 2008 2:56 pm Post subject: Reply with quote
davis21wylie2121 wrote:
There's no need to create a "new" ORtg by raising ORtg to some arbitrary exponent. ...


Ah, there's probably no 'need' for anything we do here. It just looks to some of us as though there are these 2 aspects to 'offense'; and that they haven't (already) been merged into one (usefully defined) 'rating'.

Those look like some pretty crude ranges/estimations you've got there. I'm after smooth continua, as nothing else makes sense to me. Sense is good.

Actually, reading your comments had the benefit of making me realize I had the equation backwards: I raised Usg% to a power. Whether this exponent proves to be arbitrary, or if we can find one that 'works' -- i.e., player ORtg add up to team ORtg -- remains to be seen. Balanced equations likely require a Team ORtg (and/or DRtg) factor in the individual's ORtg.

I do understand what these stats represent. Forgive me if I think we can do better.
_________________
`
36% of all statistics are wrong
Back to top
View user's profile Send private message Send e-mail
Neil Paine



Joined: 13 Oct 2005
Posts: 774
Location: Atlanta, GA

PostPosted: Thu Mar 13, 2008 3:11 pm Post subject: Reply with quote
Oy. Player ORtg, when weighted by % of total team possessions used, does add up to team ORtg.

If it ain't broke, don't fix it. Why are you so fixated on combining the two numbers into one, anyway? I find that the two offer more information when separated... Some teams need high-usage/decent efficiency guys, some teams need low-usage/high-efficiency guys. If you merge the two metrics into one number, you lose that information in translation. That's why PER is problematic in team-building -- we can't tell how a player "fits" if we don't know whether he has established a high PER on the basis of his usage or his efficiency. But with ORtg and %Poss, that's exactly what we do know. More information > less information.
Back to top
View user's profile Send private message Visit poster's website
Mountain



Joined: 13 Mar 2007
Posts: 1527


PostPosted: Thu Mar 13, 2008 3:52 pm Post subject: Reply with quote
Mike G,

I have some questions related to your original presentation:

what share of total careers since 1977 does that top 200 represent?

how do the performance means and medians compare between the ethnic groups?

how do the averages for the bottom 5-10-25% percentile groups compare?

Does the finding on one group outperforming the other in your study of the tops get repeated in reviews of the middle or the bottom? Or is the story more complicated?

I wonder what a study of undrafted player performance by ethnic group would show in comparison to the drafted. Did ethnic group performance get repeated essentially equivalently or is there a gap? What role is the front end NBA decisionmaking system / climate playing and how significant is it? Or is it mainly about the players as they come to the league, regardless of front door or side door? (genetics/culture/training experience/etc.)

The study could be extended to pay for performance too (I am not aware without checking how much and well this has already been done) . That would provide even more information and perhaps the sum of all this could provide a better basis for discussing the broad social topic you raise.
To the extent folks want to. It isn't a priority of mine but I wouldn't hide from it either.

Mountain



Joined: 13 Mar 2007
Posts: 1527


PostPosted: Thu Mar 13, 2008 4:19 pm Post subject: Reply with quote
The damage done by low usage players is determined by the team (really 5 man lineup) context, the degree of presence and ability of players who can sop up extra shots and maintain or improve or at least not fall off much.

I would not support assuming a maximum penalty for usage below the average or a penalty for low usage stronger than what Eli's recent research suggests on average unless it is in the face of specific lack of excess capacity of league average performance on a team or 5 man lineup.
Back to top
View user's profile Send private message
Mike G



Joined: 14 Jan 2005
Posts: 3623
Location: Hendersonville, NC

PostPosted: Thu Mar 13, 2008 4:29 pm Post subject: Reply with quote
davis21wylie2121 wrote:

If it ain't broke, don't fix it. ..


I didn't say it was broke. If you have a perfectly good set of cart wheels, and a perfectly good cart (without wheels), I'd just suggest we find a way to put the wheels on the cart.

You can add up the number of wheels (2+2) and determine that you have a full set. That doesn't provide anything very useful, though. You can measure the box for volume, etc; but you can't move anything in it.

Usage implies usefulness. Non-users, efficient or not, are less useful. I may be a 95% FT shooter, but if I never get to the line, it's useless. Efficiency is necessary to success, but it's not sufficient.
_________________
`
36% of all statistics are wrong
Back to top
View user's profile Send private message Send e-mail
kjb



Joined: 03 Jan 2005
Posts: 865
Location: Washington, DC

PostPosted: Fri Mar 14, 2008 9:29 am Post subject: Reply with quote
I think offensive rating is a TERRIFIC efficiency measure. But I can see the utility of what Mike's suggesting as well. Often it's good to have separate numbers for these sorts of things, and other times it's good to have things expressed in a single number. For example, efg and true shooting percentage. I don't have time to devote to coming up with the right way of combining usage and offensive rating, and frankly others with more math chops will do a better job of it anyway.
Back to top
View user's profile Send private message AIM Address Yahoo Messenger
Harold Almonte



Joined: 04 Aug 2006
Posts: 616


PostPosted: Fri Mar 14, 2008 10:06 am Post subject: Reply with quote
Please. Don't you see that the unaccurate (player mins/ Tm mins) is beggin to be eliminated as a substitution of usage in team adjusts?
Back to top
View user's profile Send private message
gabefarkas



Joined: 31 Dec 2004
Posts: 1313
Location: Durham, NC

PostPosted: Sun Mar 23, 2008 9:14 pm Post subject: Reply with quote
Mike G wrote:
Usage implies usefulness. Non-users, efficient or not, are less useful. I may be a 95% FT shooter, but if I never get to the line, it's useless. Efficiency is necessary to success, but it's not sufficient.

Short answer: So what?

Longer answer: Although it's likely true, why does this mean you need to combine them into one number somehow? How does that help the situation? I believe most people capable of perusing this here message board are also able to process more than one piece of information at a time. That is, humans are more complex than single-bit machines; it's a skill that is already well-developed by the time we've learned how to "carry the one" in grade school. So, what's wrong with looking at two numbers at the same time?
Back to top
View user's profile Send private message Send e-mail AIM Address

94by50



Joined: 01 Jan 2006
Posts: 499
Location: Phoenix

PostPosted: Sun Mar 23, 2008 10:30 pm Post subject: Reply with quote
gabefarkas wrote:
Longer answer: Although it's likely true, why does this mean you need to combine them into one number somehow? How does that help the situation? I believe most people capable of perusing this here message board are also able to process more than one piece of information at a time. That is, humans are more complex than single-bit machines; it's a skill that is already well-developed by the time we've learned how to "carry the one" in grade school. So, what's wrong with looking at two numbers at the same time?

Correct me if I'm wrong, but isn't the real problem that we don't quite know what combinations of usage and efficiency are optimal? For example, dealing with individual players, a player with an offensive rating of 105 using 30% of the offense, and a player with an offensive rating of 120 using 12% of the offense... which one's more valuable? This doesn't even get us into 5 players functioning as one unit and how their efficiency/usage combinations can be optimized.
Back to top
View user's profile Send private message
mikedc



Joined: 11 Aug 2005
Posts: 5
Location: Indianapolis, IN

PostPosted: Tue Apr 01, 2008 2:44 pm Post subject: Reply with quote
That's interesting stuff and an interesting thread in general. If I have the time, it sparked interest in a new research project for me. Unfortunately that'll have to take its place in line behind the other five or six hundred research ideas I like but haven't gotten around to yet Smile

I think there's some truth to what Chocolate Thunder is saying, but I actually draw the opposite conclusion.

Basically what I see in the numbers is:
1. whites over-represented in most categories
2. whites wildly ahead in things like 3 point shooting
3. whites very under-represented in the usage rate statistic.

Being an economist, my first guess is this these differences can all be accounted for without resorting to explaining them as physical differences due to race or any sort of innate racial differences in "taste" the way Dawkins seems to.

To do that, I'd start at the opposite point from Dawkins. The "white" game is too individualistic and the "black" game is too socialized.

Consider it in terms of trying to develop a set of skills. I would guess that if conditions for an athletic white kid and an athletic black kid are different, they're largely different like this.

1. Your stereotypical "black kid" lives in a city and is less likely to have his own basketball hoop. To develop his basketball skills, this requires him to play with other guys; whomever is at the court. This is going to teach the better players to use the ball more, and it's going to lead them to focus on things that win them games. If you're playing with other kids and you do nothing but stand around and wait for someone to pass you the ball so you can hoist up a three pointer, you're never gonna touch the ball. So the skills that get reinforced and practiced in that environment are things like controlling the ball and shooting.

2. Your "stereotypical white kid" might live in the suburbs or country, and might have his own basketball hoop. Even if he doesn't, there are less other kids around, so he's going to spend more time developing his skills alone. Bouncing a ball against his garage door to simulate passes, practicing his jumper, collecting the misses etc. On the other hand, there's not that same level of competition for "ball time" you find in a game setting.

1+2 explain why you see white kids coming to the table with superior 3 point skills and black kids coming to table with superior "usage" skills.

3. The fact that whites are somewhat over-proportionally represented amongst the best is likely due to a broader manifestation of the same principle of practice and specialization. When our stereotypical black kid turns out to be a good athlete, he might focus on basketball or football and develop one of those skills because they're the most obvious payoffs for him. When our stereotypical white kid is a good athlete, my guess is there are still quite a bit more things he might specialize in. Tennis, golf, baseball, hockey and perhaps non-athletic pursuits as well, all compete for time.

This leads to over-representation among the statistical leaders at the NBA level because, in order to advance actually advance to that level, the average white player must have dispayed a higher aptitude for it in the first place to give up his other opportunities and specialize in basketball in the first place. Only the most skilled and dedicated white players go on to the highest levels of basketball, and thus, they tend to be among the best.

This is true of black kids too, of course, but the number of other things that good black athletes end up specializing in and dedicating their time to ends up being smaller, so a higher proportion of the "potential basketball players" become "actual basketball players".

I spose I could be completely wrong, and I'd like to find some good tests of my theory here, but I do think it's a logically sound explanation at least.
_________________
Sportstwo.com
Back to top
View user's profile Send private message Send e-mail Visit poster's website AIM Address Yahoo Messenger MSN Messenger
Chicago76



Joined: 06 Nov 2005
Posts: 98


PostPosted: Tue Apr 01, 2008 11:01 pm Post subject: Reply with quote
I agree with a lot of the sentiment behing Mikedc's post. Namely, that growing up in an environment where you had to win amongst a group of guys to literally keep the court could produce a different style of ball than a solitary shooter working on his jump shot alone into the night.

I'm going to throw stats out the window and speak anecdotally about my home town of Indianapolis. There was a court in Indianapolis that is probably about as historic and competitive as any in the country not named Rucker. This court sat just west of downtown where IUPUI is today and was called the "Dust Bowl". This court was used by African Americans in a more crowded part of town where resources were few, playing time was scarce, and players needed to learn to fend for themselves to stay on the court. Due to scarcity of court time, the biggest, most physically gifted players learned to play on this court and they had to take on roles they wouldn't need to if they were a 6'5" kid practicing 12 footers in the country. This court produced two players in particular who were prototypes of the modern NBA player. The first player is universally known: Oscar Robertson, the first big point guard who could also rebound quite well. The second didn't have as much success due to his own work ethic, but was the predecessor to a Chris Webber, Kevin Garnett type of power forward who could put the ball on the floor, rebound, and pass well--George McGinnis.

There is an interesting passage here that contains a lot of truth and provides a lot of historical context, once you get past some of the stereotypes:

http://www.undertheboards.com/hoosiers.htm

I suspect this scenario isn't unique to Indiana.

Coincidentally, if you were wondering about that white rural "other" Hoosier who was a pretty good shooter, passer, and rebounder for a small forward. He grew up in a triangle of the state that continued to play a pretty free-flowing game (Louisville to Evansville north a bit toward Bloomington) due to geographic isolation and the fact that many of the small towns in this part of the state integrated historically early (Washington HS being one).
Back to top
View user's profile Send private message
Harold Almonte



Joined: 04 Aug 2006
Posts: 616


PostPosted: Thu Apr 03, 2008 9:47 am Post subject: Reply with quote
Mikedc. You put a lot of weight on your cultural, economical and geographic environment theory of development-example. That can be put validly in the mix, but that does not define the two game tendencies as you explained it.

Would you say that the Princenton offense, the extreme and caricaturisthical episthome of "white" game, is individualistic?, and the one on one offense, the extreme and caricaturisthical episthome of "black" game, is social?

Don't underrate Biology, it has a lot of weight.
Back to top
View user's profile Send private message
mikedc



Joined: 11 Aug 2005
Posts: 5
Location: Indianapolis, IN

PostPosted: Thu Apr 03, 2008 3:24 pm Post subject: Reply with quote
Harold Almonte wrote:
Mikedc. You put a lot of weight on your cultural, economical and geographic environment theory of development-example. That can be put validly in the mix, but that does not define the two game tendencies as you explained it.

Would you say that the Princenton offense, the extreme and caricaturisthical episthome of "white" game, is individualistic?, and the one on one offense, the extreme and caricaturisthical episthome of "black" game, is social?

Don't underrate Biology, it has a lot of weight.


On biology, contrary to the popular perceptions I've always thought that argument worked against the predominance of blacks in American sports.

Let me just acknowledge up front that I'm not a biologist so this is purely a conjecture based on my knowledge of statistics:

Even if you were to say that blacks were better physical specimens on average than whites, on average there's no cartoonish sort of obvious differences. Pick the average black guy off the street and the average white guy off the street and physically they bear a much closer resemblence to each other than they do to Lebron James.

I expect (but do not know) that both white and white physical abilities are basically normally distributed. Lebron, for example, is certainly somewhere in the far right tail of the distribution of physical gifts amongst both blacks and everyone.

If this is the case, there must be more "White Lebrons" out there than there are "Black Lebrons", because there are roughly 9 white guys out there for every 1 black guy. Again, if we follow a normal distribution, there are going to be more whites that are good physical specimens, in absolute terms than blacks. Even if blacks are slightly better on average I'd think this would still be the case.

If this is the case, I think the ultimate answer cannot be based on physical differences; I think there's actually a much bigger pool of physically gifted whites to draw on than physically gifted blacks. Hence, the difference must come in what physically gifted blacks and whites end up doing. Physically gifted blacks end up playing basketball at a much higher rate than physically gifted whites to.

-------

Quote:
Would you say that the Princenton offense, the extreme and caricaturisthical episthome of "white" game, is individualistic?, and the one on one offense, the extreme and caricaturisthical episthome of "black" game, is social?


To get all metaphysical, neither game can literally be "individualistic" since it's being played with others Very Happy I'll admit to sort of playing on words there to get at the underlying idea of "scarcity of ball or court time". As Chicago76 pointed out more clearly than I, a kid who develops his skills in a high-stakes but still disorganized environment is going to develop them very differently than a guy who doesn't have to fight for the resources and/or learns in an organized environment.

By individualistic vs. social, I simply meant that. Ironically, being part of a crowd can teach us we have to stand out and fight for the ball, while being alone with the ball can teach us how to play a team game.
_________________
Sportstwo.com
Back to top
View user's profile Send private message Send e-mail Visit poster's website AIM Address Yahoo Messenger MSN Messenger
Harold Almonte



Joined: 04 Aug 2006
Posts: 616


PostPosted: Fri Apr 04, 2008 11:04 am Post subject: Reply with quote
Again I partially agree with you on the influence of cultural and economics on skills development (and probably even on Darwinist biological evolution), the need is the mother of inventions. It happens that psycologists say that there are seven types of intelligence, and nobody has all of them developed at the same level, and some people are, born, or grow, or develop better or different than others at some of them. But, the "physical gifts" between races, tend to be "supposedly" differentiated by some evolutioned geneticals that supposedly affect muscular cell's methabolism. Experts (I'm not one of them) speculate that even in an even ratial pool of "physically gifted" sportsmen, blacks will have an advantage at sports which highly employ athletical body moves, while white will advantage at different things.

Yes, a team sport is much more than basic athletic routines, but the athletical advantage makes the game strategically easier, and while disorganized and not efficient in the use of own energy and strenght, it's very efficient in the exploit of opponents's mechanical weaknesses, in a game with limits of time for execution and acomplishments, and that it has evolutioned almost in a 3 on 3 of individual matchups.

P.D.
The fabulous dream of the Memphis's one on one-drive and dribble-athletical game, was truncated by a short lack-of-strategy lapse, and by true shooting skills, in the moments it mattered most.

Mark



Joined: 20 Aug 2005
Posts: 807


PostPosted: Wed Apr 25, 2007 2:22 pm Post subject: Splits of data by usage levels and more Reply with quote
There are a lot of splits available for player data- by position, by minutes. by starter/sub., against good and bad teams, etc. One more that might be useful: by usage. Take game data and split into groups of usage over 20, 15-20. 10-15 and below 10. It would aid discussion about how to split shots on a team and how much a player has shown he can handle. Would hoopstats, basketball-reference or 82games or ESPN or any others have interest in doing this? Has any analyst done this for their home team or have interest in doing so and sharing it?

Since usage greatly affects PER and many other metrics it might be interesting to divide lists and show just top 2 usage on team guys on one list and rest of team guys on another so that you see the best/average/worst of each peer group. And then maybe split the list further by above .500 team and below.

Heard Chad Ford and Henry Abott podcast from yesterday talking about stats in context prompted this thought that splitting lists by various context criteria could be worked more.Any comments or further thoughts?

Last edited by Mark on Wed Apr 25, 2007 8:09 pm; edited 3 times in total
Back to top
View user's profile Send private message
Ben F.



Joined: 07 Mar 2005
Posts: 391


PostPosted: Wed Apr 25, 2007 2:57 pm Post subject: Reply with quote
That's an interesting idea, and it certainly would be good to see at a glance. The only thing I'd say is you still won't be able to draw conclusions from it, since as has been said before, often times players who have good matchups or seem to be making their shots will get more shots thus increasing usage because of efficiency, so it's not the greatest test case.
Back to top
View user's profile Send private message
Mark



Joined: 20 Aug 2005
Posts: 807


PostPosted: Wed Apr 25, 2007 3:24 pm Post subject: Reply with quote
I agree with your caution about drawing conclusions as is also the case with splits by minutes.
Back to top
View user's profile Send private message
Harold Almonte



Joined: 04 Aug 2006
Posts: 616


PostPosted: Thu Apr 26, 2007 8:59 am Post subject: Reply with quote
Good idea. As we know that usage is just a measure of offensive power (not the better use of the power), we can get a piramidal look of the NBA power. It would be fair to add a defensive usage too.

P.D. - Power=Money

ErichDoerr



Joined: 06 Jul 2008
Posts: 15


PostPosted: Wed Apr 15, 2009 9:06 pm Post subject: NBA Playoffs simulator Reply with quote
I am working on an Excel spreadsheet to estimate playoff odds and track them throughout the playoffs. The worksheet is linked below:

http://xlssports.googlepages.com/xlsSpo ... ayoffs.xls

I would appreciate any feedback you may have and am working on improvements.
Back to top
View user's profile Send private message Send e-mail Visit poster's website
ErichDoerr



Joined: 06 Jul 2008
Posts: 15


PostPosted: Wed Apr 15, 2009 11:53 pm Post subject: Reply with quote
I have posted this file at http://www.xlssports.com/2009/04/nba-pl ... lator.html

Updates made in the last hour include corrections to HCA and calculations to account for results tracking from bracket tab.
Back to top
View user's profile Send private message Send e-mail Visit poster's website
Serhat Ugur (hoopseng)



Joined: 13 Oct 2006
Posts: 209
Location: Basketball Research

PostPosted: Thu Apr 16, 2009 3:27 am Post subject: Reply with quote
Erich,
It looks like Cavs~Lakers matchup in the finals is a lock.

What made me confused is your playoff format table, is it correct?

H: home?
A: away?

If so,

Till the finals: it should be H-H-A-A-H-A-H
Finals: H-H-A-A-A-H-H
_________________
http://www.nbastuffer.com
Back to top
View user's profile Send private message Visit poster's website
Ben



Joined: 13 Jan 2005
Posts: 266
Location: Iowa City

PostPosted: Thu Apr 16, 2009 10:34 am Post subject: Reply with quote
Looks very nice. And yes, it does look like the playoff format table needs correction as hoopseng mentioned.
Back to top
View user's profile Send private message
ErichDoerr



Joined: 06 Jul 2008
Posts: 15


PostPosted: Thu Apr 16, 2009 11:31 am Post subject: Reply with quote
Thanks guys,
I updated the playoff format as mentioned.

Cavs/Lakers is more likely now that Garnett looks out, Ginobili is out, and Bynum is back.

I'm sure somebody here can come up with better Pythag%'s than I use, so please advise if you have any suggestions
Back to top
View user's profile Send private message Send e-mail Visit poster's website
Ben



Joined: 13 Jan 2005
Posts: 266
Location: Iowa City

PostPosted: Fri Apr 17, 2009 1:43 pm Post subject: Reply with quote
I was messing around to see the effect of home court and found something odd. If you give Cleveland 40 for HCA, then Detroit has 97% chance of winning round 1. Something's off there - I haven't examined it further to find the cause.
Back to top
View user's profile Send private message
ErichDoerr



Joined: 06 Jul 2008
Posts: 15


PostPosted: Sat Apr 18, 2009 9:24 am Post subject: Reply with quote
Ben,
Thanks again.

I have found the offending formula and believe it has been addressed. I have also updated the H/R Pythags for the final season values I found via B-R.com

Updated file uploaded to http://xlssports.googlepages.com/xlsSpo ... ayoffs.xls
Back to top
View user's profile Send private message Send e-mail Visit poster's website
ErichDoerr



Joined: 06 Jul 2008
Posts: 15


PostPosted: Sat Apr 18, 2009 10:04 am Post subject: Reply with quote
If I were to use the B-R data, some injury adjustments, and make the C-List for
Henry Abbott's Stat Geek Smackdown, I would submit my picks as follows (odds from linked worksheet):

Pick Winner R1Win 4-0 4-1 4-2 4-3
Win 4-1 CLE .969 .313 .412 .139 .104
Win 4-1 BOS .903 .181 .370 .163 .188
Win 4-1 ORL .911 .175 .382 .159 .195
Win 4-3 ATL .688 .049 .222 .118 .299
Win 4-1 LAL .901 .126 .374 .153 .248
Win 4-3 DEN .683 .062 .226 .129 .267
Win 4-3 SAS .623 .059 .196 .127 .242
Win 4-3 POR .697 .035 .211 .105 .347

Mike G



Joined: 14 Jan 2005
Posts: 3604
Location: Hendersonville, NC

PostPosted: Tue May 19, 2009 8:23 am Post subject: Reply with quote
2-round eWins leaders. Column at right is eWins per 484 minutes; 1.00 is the average at this level.
Code:
eW per36 rates tm G Eff% Sco Reb Ast Stl TO Blk T e484
2.75 Howard,Dwight Orl 12 .607 22.5 17.8 1.2 .6 2.6 2.7 42.7 2.26
2.45 James,Lebron Cle 8 .625 43.8 10.9 7.2 1.9 1.5 .6 65.7 3.89
2.36 Bryant,Kobe LAL 12 .540 29.7 5.3 4.3 2.0 2.4 .9 41.8 2.19
1.87 Rondo,Rajon Bos 14 .460 14.0 9.3 7.6 2.2 2.4 .2 33.7 1.62
1.83 Nowitzki,Dirk Dal 10 .619 26.2 10.4 2.5 .8 2.2 .8 39.4 2.02

1.78 Gasol,Pau LAL 12 .575 20.8 11.5 2.1 .7 2.1 1.7 35.6 1.75
1.73 Anthony,Carm Den 10 .566 28.3 6.9 3.9 1.9 1.9 .8 41.6 2.18
1.70 Lewis,Rashard Orl 13 .543 20.2 6.2 2.8 .9 2.0 .8 30.0 1.35
1.53 Billups,Chau Den 10 .697 26.4 4.1 6.7 1.2 1.7 .3 39.3 2.01
1.52 Wade,Dwyane Mia 6 .560 30.8 5.5 5.0 .8 3.2 1.4 41.7 2.18

1.45 Pierce,Paul Bos 14 .541 20.0 5.8 2.5 1.0 2.2 .3 27.9 1.21
1.42 Odom,Lamar LAL 12 .600 17.8 13.1 2.6 .7 2.3 1.8 34.6 1.68
1.34 Perkins,Kend Bos 14 .585 12.8 12.5 1.3 .4 2.4 2.7 28.0 1.21
1.32 Ming,Yao Hou 9 .618 20.4 11.9 .9 .5 1.9 1.4 33.3 1.59
1.29 Smith,Josh Atl 11 .483 17.2 8.2 2.1 1.1 1.9 1.5 29.2 1.30

eW per36 rates tm G Eff% Sco Reb Ast Stl TO Blk T e484
1.23 Scola,Luis Hou 13 .518 17.0 10.0 1.8 .6 1.9 .3 27.9 1.21
1.22 Roy,Brandon Por 6 .552 28.2 5.3 2.4 1.3 2.1 1.1 37.3 1.87
.99 Brooks,Aaron Hou 13 .565 20.0 3.0 3.3 .4 2.4 .3 25.1 1.00
.91 Davis,Glen Bos 14 .527 16.1 6.2 1.6 1.3 1.5 .6 24.9 .99
.91 Parker,Tony SAS 5 .579 31.6 4.6 5.8 1.2 4.3 .2 40.3 2.09

.89 Artest,Ron Hou 13 .468 15.1 4.5 3.7 1.1 2.4 .2 23.0 .86
.86 Ariza,Trevor LAL 12 .642 17.3 5.2 3.7 1.7 2.4 .6 27.3 1.17
.85 Bibby,Mike Atl 11 .620 16.4 4.0 4.2 1.0 2.3 .2 24.4 .96
.83 Johnson,Joe Atl 11 .472 15.6 4.7 3.2 1.2 2.7 .0 22.7 .84
.83 Kidd,Jason Dal 10 .613 11.3 6.1 4.9 2.1 1.9 .3 24.8 .98

.81 Howard,Josh Dal 10 .518 18.8 7.1 1.4 1.1 2.7 .5 26.5 1.11
.80 Smith,J.R. Den 10 .601 24.0 5.0 3.1 1.5 1.7 .3 33.2 1.58
.80 Boozer,Carlos Uta 5 .575 19.6 13.6 1.8 1.6 3.0 .4 34.3 1.66
.78 Turkoglu,Hedo Orl 13 .524 15.5 4.1 4.2 .8 2.6 .2 22.9 .85
.77 Hilario,Nene Den 10 .571 15.1 9.0 2.6 1.5 2.5 .9 27.5 1.18

eW per36 rates tm G Eff% Sco Reb Ast Stl TO Blk T e484
.75 Iguodala,Andr Phl 6 .533 18.1 5.8 4.6 1.5 3.3 .0 27.9 1.20
.75 Duncan,Tim SAS 5 .544 23.3 9.9 3.0 .7 1.6 1.4 37.9 1.92
.74 Allen,Ray Bos 14 .545 17.3 3.9 2.2 1.0 1.7 .3 23.7 .91
.71 Miller,Andre Phl 6 .538 18.7 6.0 3.8 1.0 2.7 .1 27.8 1.20
.70 Williams,Mo Cle 8 .551 19.8 3.1 5.2 .5 2.3 .1 27.6 1.18

.69 Ilgauskas,Zyd Cle 8 .473 16.6 10.3 2.0 .5 1.3 1.5 30.4 1.38
.69 Alston,Rafer Orl 12 .471 15.0 3.2 5.1 1.8 1.9 .2 25.3 1.02
.68 Aldridge,Lamar Por 6 .513 18.5 7.7 1.1 .5 1.6 1.4 28.1 1.22
.67 West,Delonte Cle 8 .567 16.2 4.2 4.5 1.0 2.4 .2 25.1 1.01

_________________
`
36% of all statistics are wrong
Back to top
View user's profile Send private message Send e-mail
Statman



Joined: 20 Feb 2005
Posts: 242
Location: Arlington, Texas

PostPosted: Wed May 20, 2009 2:00 am Post subject: Reply with quote
So looking at e484, Lebron & a completely average player (say, Delonte West) combined are performing better than the two other best players in the playoffs (Dwight Howard & Kobe) combined.

Man, Lebron totally doesn't suck.
_________________
Dan

My current national college player rankings (and other stuff):
http://www.pointguardu.com/f136/statman ... post355594
Back to top
View user's profile Send private message Send e-mail
Mike G



Joined: 14 Jan 2005
Posts: 3604
Location: Hendersonville, NC

PostPosted: Thu May 28, 2009 7:31 am Post subject: Reply with quote
Mike G wrote:
After 4 games for everyone, ...
.. the breakdown by position:
Code:
pos Min eW %Min e484
C 2957 6.39 .190 1.05
PF 2997 6.82 .193 1.10
SF 3283 6.00 .211 0.88
SG 2584 4.41 .166 0.83
PG 3735 8.38 .240 1.09

all 15556 32.00 1.000 1.00


Thru 2 rounds, PG minutes have remained high, but production apparently has dropped much.
Code:
pos Min %Min eW e484
C 6154 .190 13.50 1.04
PF 6172 .191 14.56 1.12
SF 6591 .204 13.25 .95
SG 5454 .169 10.86 .94
PG 7969 .246 15.84 .94

all 32341 1.00 68.00 1.00
Production from SF and SG soared in round 2. Odd.
_________________
`
36% of all statistics are wrong
Back to top
View user's profile Send private message Send e-mail
eyriq



Joined: 04 Jun 2008
Posts: 54
Location: Orlando

PostPosted: Thu May 28, 2009 9:13 am Post subject: Reply with quote
Statman wrote:
So looking at e484, Lebron & a completely average player (say, Delonte West) combined are performing better than the two other best players in the playoffs (Dwight Howard & Kobe) combined.

Man, Lebron totally doesn't suck.


That is horrifying.
Back to top
View user's profile Send private message
Mike G



Joined: 14 Jan 2005
Posts: 3604
Location: Hendersonville, NC

PostPosted: Tue Jun 02, 2009 7:10 pm Post subject: Reply with quote
eWins leaders thru the conference finals.
Code:
eW per36 rates tm G Eff% Sco Reb Ast Stl TO Blk T e484
4.05 James,Lebron Cle 14 .600 39.1 9.8 6.5 1.5 2.4 .8 56.9 3.43
3.60 Bryant,Kobe LAL 18 .565 33.3 5.6 4.3 1.6 2.3 .7 44.3 2.48
3.46 Howard,Dwight Orl 18 .629 25.3 16.9 1.8 .6 2.5 2.2 44.0 2.45
2.72 Gasol,Pau LAL 18 .600 21.0 11.8 2.2 .8 2.0 1.9 36.0 1.85
2.41 Anthony,Carmelo Den 16 .552 27.0 6.1 3.4 1.7 2.3 .6 37.0 1.93

2.06 Billups,Chauncey Den 16 .646 22.8 4.0 5.6 1.2 1.9 .2 33.3 1.65
2.03 Lewis,Rashard Orl 19 .565 19.5 6.2 2.5 1.0 1.9 .7 28.4 1.28
1.88 Rondo,Rajon Bos 14 .460 14.0 9.5 7.2 2.2 2.4 .2 32.5 1.60
1.74 Odom,Lamar LAL 18 .568 16.6 12.4 2.3 .7 1.9 1.7 31.9 1.55
1.72 Nowitzki,Dirk Dal 10 .619 27.1 10.6 2.6 .9 2.2 .8 39.9 2.15

1.44 Turkoglu,Hedo Orl 19 .522 15.6 5.0 5.2 .7 2.6 .1 24.2 .97
1.40 Ariza,Trevor LAL 18 .657 19.7 5.4 3.1 2.0 2.4 .7 28.8 1.31
1.38 Pierce,Paul Bos 14 .541 20.3 5.9 2.4 1.0 2.2 .3 27.5 1.21
1.32 Wade,Dwyane Mia 7 .554 31.3 5.1 5.3 .8 3.3 1.4 41.4 2.26
1.29 Perkins,Kendrick Bos 14 .585 13.1 12.8 1.2 .4 2.4 2.7 27.7 1.23

1.18 Alston,Rafer Orl 18 .477 14.5 3.3 5.3 1.9 2.1 .3 24.5 .99
1.16 Hilario,Nene Den 16 .571 14.1 9.6 2.6 1.5 2.3 .7 26.2 1.12
1.09 Ming,Yao Hou 9 .618 21.1 12.0 .9 .5 2.0 1.4 33.4 1.66
1.09 Smith,Josh Atl 11 .483 17.3 8.3 1.9 1.1 1.9 1.5 28.5 1.29
1.09 Ilgauskas,Zydrun Cle 14 .476 14.2 12.4 1.6 .5 .9 1.2 28.9 1.32

1.05 Williams,Mo Cle 14 .524 17.5 3.7 4.0 .7 2.1 .1 24.0 .96
1.05 Scola,Luis Hou 13 .518 17.3 10.2 1.7 .6 2.0 .3 27.5 1.21
1.03 Davis,Glen Bos 14 .527 16.4 6.3 1.5 1.3 1.5 .6 24.5 .99
1.03 Allen,Ray Bos 14 .545 17.6 4.0 2.0 1.0 1.7 .3 23.4 .91
1.00 Smith,J.R. Den 16 .551 19.3 4.6 3.0 1.4 2.1 .3 26.5 1.15


And a Finals preview:
Code:
e484 LA Lakers Min Eff% Sco Reb Ast Stl TO Blk T
2.48 Bryant,Kobe 39 .565 33.3 5.6 4.3 1.6 2.3 .7 44.3
1.85 Gasol,Pau 39 .600 21.0 11.8 2.2 .8 2.0 1.9 36.0
1.55 Odom,Lamar 30 .568 16.6 12.4 2.3 .7 1.9 1.7 31.9
1.31 Ariza,Trevor 29 .657 19.7 5.4 3.1 2.0 2.4 .7 28.8
1.10 Bynum,Andrew 16 .514 15.8 9.1 .8 .8 2.0 2.3 25.9

.45 Fisher,Derek 24 .455 11.0 3.0 3.3 1.3 1.9 .1 17.0
.73 Brown,Shannon 14 .554 15.7 3.5 1.6 1.3 1.3 .3 21.0
1.07 Farmar,Jordan 11 .503 14.3 4.1 4.8 1.2 2.5 .5 23.2
.69 Walton,Luke 14 .430 8.0 6.2 4.8 1.7 2.4 .4 19.5
.36 Vujacic,Sasha 12 .392 9.2 5.0 1.4 1.5 1.3 .6 16.1

e484 Orlando Magic Min Eff% Sco Reb Ast Stl TO Blk T
2.59 Howard,Dwight 36 .629 25.3 16.9 1.8 .6 2.5 2.2 44.0
1.28 Lewis,Rashard 40 .565 19.5 6.2 2.5 1.0 1.9 .7 28.4
.97 Turkoglu,Hedo 38 .522 15.6 5.0 5.2 .7 2.6 .1 24.2
1.05 Alston,Rafer 30 .477 14.5 3.3 5.3 1.9 2.1 .3 24.5
.81 Pietrus,Mickael 25 .624 16.7 4.2 .9 1.1 1.4 .9 22.0

1.06 Gortat,Marcin 11 .721 12.7 11.2 .5 1.1 1.3 1.6 25.4
.54 Lee,Courtney 24 .526 11.8 2.9 2.2 1.3 1.4 .2 17.4
.55 Johnson,Anthony 14 .436 9.3 3.6 4.9 1.3 1.5 .0 18.7
.86 Redick,J.J. 13 .532 11.2 2.8 3.4 1.0 .7 .1 18.5
.69 Battie,Tony 5 .479 12.0 6.6 .0 .0 .4 .4 18.1

The e484 in these lists is calculated a couple of different ways. No profound inconsistencies.

The Magics have their work cut out for them, as the Lakers are loaded.
Unless, in fact, these Orlando stats came against superior competition.
_________________
`
36% of all statistics are wrong
Back to top
View user's profile Send private message Send e-mail
Mike G



Joined: 14 Jan 2005
Posts: 3604
Location: Hendersonville, NC

PostPosted: Mon Jun 15, 2009 11:22 am Post subject: Reply with quote
First-pass eWins allocation for the 2009 playoffs.
Some reduction from previous post, from the best players' eWins, due to lower 'replacement' (zero eW) level. Still a work in progress.

Code:
eW per36 rates tm eG Eff% Sco Reb Ast T e484
4.51 Bryant,Kobe LAL 25.5 .553 32.4 5.5 4.6 43.7 2.37
4.01 Howard,Dwight Orl 24.8 .611 22.3 16.3 1.8 40.9 2.17
3.90 James,Lebron Cle 15.9 .600 38.7 9.8 6.5 56.6 3.30
3.34 Gasol,Pau LAL 25.6 .607 20.7 11.1 2.1 35.2 1.76
2.47 Lewis,Rashard Orl 27.1 .560 18.6 6.4 2.7 27.8 1.22

2.34 Anthony,Carmelo Den 16.8 .552 26.8 6.1 3.3 36.8 1.88
2.12 Odom,Lamar LAL 19.8 .575 16.8 11.6 1.9 30.8 1.44
2.01 Billups,Chauncey Den 16.8 .646 22.6 4.0 5.6 33.1 1.61
1.88 Turkoglu,Hedo Orl 25.6 .536 16.0 4.9 4.8 24.4 .98
1.85 Rondo,Rajon Bos 15.9 .460 14.0 9.4 7.2 32.5 1.56

1.67 Nowitzki,Dirk Dal 10.8 .619 26.8 10.6 2.5 39.7 2.09
1.60 Ariza,Trevor LAL 19.5 .599 16.9 5.8 2.6 26.1 1.10
1.37 Pierce,Paul Bos 15.3 .541 20.2 5.9 2.3 27.5 1.21
1.34 Alston,Rafer Orl 20.1 .467 13.9 3.2 5.0 23.2 .89
1.29 Wade,Dwyane Mia 7.8 .554 31.1 5.1 5.3 41.4 2.21

eW per36 rates tm eG Eff% Sco Reb Ast T e484
1.27 Perkins,Kendrick Bos 14.1 .585 13.0 12.8 1.2 27.6 1.21
1.15 Hilario,Nene Den 14.0 .571 14.0 9.5 2.6 26.0 1.10
1.07 Smith,Josh Atl 11.3 .483 17.3 8.2 1.9 28.4 1.27
1.07 Ilgauskas,Zydrun Cle 11.1 .476 14.2 12.3 1.6 28.8 1.30
1.07 Ming,Yao Hou 8.8 .618 20.9 12.0 .9 33.3 1.63

1.07 Williams,Mo Cle 14.8 .524 17.5 3.7 4.0 24.2 .97
1.05 Scola,Luis Hou 11.6 .518 17.2 10.2 1.7 27.6 1.21
1.03 Allen,Ray Bos 15.2 .545 17.6 4.0 2.0 23.4 .91
1.02 Davis,Glen Bos 13.9 .527 16.3 6.3 1.5 24.4 .98
1.02 Smith,J.R. Den 11.8 .551 19.4 4.6 3.1 26.9 1.16
Minutes and games are compressed into a single column : equivalent (36-minute) 'games'; that is, Min/36.
_________________
`
36% of all statistics are wrong
Back to top
View user's profile Send private message Send e-mail
eyriq



Joined: 04 Jun 2008
Posts: 54
Location: Orlando

PostPosted: Tue Jun 16, 2009 9:48 pm Post subject: Reply with quote
Hey Mike, love the numbers. Could you fancy me and explain what the "T" does and means? Thanks.
Back to top
View user's profile Send private message
Mike G



Joined: 14 Jan 2005
Posts: 3604
Location: Hendersonville, NC

PostPosted: Wed Jun 17, 2009 9:00 am Post subject: Reply with quote
T is just indicative of a Total of the other columns (Sco, Reb, Ast, Stl, TO, Blk, PF). Ranking players by 'T' is the same as their ranking by 'e484', so it's kind of redundant. EWins is = (T - Tr)*Min/x , where Tr is the T rate of a 'replacement level player', X is a 'variable constant'.

Because player eWins sum to team eWins, the subsidiary columns (Sco, etc) are necessarily weighted for best fit between teams' eWins and their Pythagorean (pt-diff) expected wins.

As such, the T rate is sort of a temporary bridge between productivity rates and equivalent-Wins (eW). I've omitted it from such posts and was asked to bring it back. It has some reference value, I guess. Broadly, a T of 40 is a superstar, 35 is solid allstar, 30 is a minor star, 25 is solid starter, 20 is solid rotation player, 15 is a role/bit player. You might call 45 a megastar, 50 an ultrastar?

I've found that T rates in playoffs tend to average 90-95% of what they were in the regular season. This after the usual adjustments for scoring/rebounding pace. Stars tend to drop off less and role players more.
_________________
`
36% of all statistics are wrong
Back to top
View user's profile Send private message Send e-mail
eyriq



Joined: 04 Jun 2008
Posts: 54
Location: Orlando

PostPosted: Wed Jun 17, 2009 10:13 am Post subject: Reply with quote
Ok, great! Thank you. Any reason why you prefer this method over other box score systems, or probably more appropriately phrased is why should someone approve of this method over some of the other methods out there like PER? One advantage that I can think of is your end product, eWins. I think that provides a clear and easily understood symbology with built in applicability.
Back to top
View user's profile Send private message
Mike G



Joined: 14 Jan 2005
Posts: 3604
Location: Hendersonville, NC

PostPosted: Wed Jun 17, 2009 10:32 am Post subject: Reply with quote
I have a 'scoring' rate that melds production and accuracy. I don't see how good a scorer someone might be by looking at PER, or at ORtg and/or Usg%, unless I do another calculation.

In playoffs even more than seasons, a team's success (like 0-4) can swamp the contributions of an individual, in those other systems. eWins apportionment falls in between the ratio of wins and the ratio of points scored (team/opponent).

This system also estimates % of points unassisted, % of minutes vs starters; Reb rate is scaled more to opponent Reb than to team's; I'm sure there are other things.

All these standardizations attempt to mathematically place the individual on an average team. In playoffs, the standards are higher; the best players can still get their normal share of eWins, though.
_________________
`
36% of all statistics are wrong
Back to top
View user's profile Send private message Send e-mail
eyriq



Joined: 04 Jun 2008
Posts: 54
Location: Orlando

PostPosted: Wed Jun 17, 2009 11:47 am Post subject: Reply with quote
Thanks again, I just realized you are probably the same Mike from Hoopsanalyst, a site I visit quite frequently. Cool, this place is freaking great, lol.

Mike G



Joined: 14 Jan 2005
Posts: 3553
Location: Hendersonville, NC

PostPosted: Sat May 01, 2010 2:19 pm Post subject: 2010 Playoffs Summaries Reply with quote
EDIT : Final 1st Round eWins/rates, top 50-some.
Code:
eW/G eW per36 rates tm mpg Eff% Sco Reb Ast e480
.292 1.46 James,Lebron Cle 41 .652 34.6 8.7 6.7 3.47
.263 1.32 Wade,Dwyane Mia 42 .638 34.9 5.6 5.2 3.05
.229 1.37 Richardson,Jason Phx 32 .665 35.1 9.0 1.1 3.40
.215 .86 Nelson,Jameer Orl 36 .633 32.1 3.4 5.4 2.86
.192 1.15 Nowitzki,Dirk Dal 39 .630 32.5 8.0 2.7 2.38

.185 1.11 Westbrook,Russel Okl 36 .559 23.9 6.3 6.4 2.50
.183 1.10 Gasol,Pau LA 36 .571 20.8 12.5 3.4 2.43
.175 1.05 Williams,Deron Uta 40 .633 25.5 2.7 8.7 2.11
.171 1.02 Anthony,Carmelo Den 42 .553 25.7 8.4 2.3 1.96
.160 .96 Boozer,Carlos Uta 41 .577 19.4 13.3 2.4 1.88

.155 .77 Rondo,Rajon Bos 42 .493 14.3 6.1 9.0 1.78
.150 .90 Durant,Kevin Okl 39 .487 24.6 7.4 2.3 1.86
.140 .70 Allen,Ray Bos 35 .678 26.9 3.0 3.6 1.92
.138 .83 Bryant,Kobe LA 37 .498 25.6 4.1 3.9 1.81
.136 .55 Wallace,Gerald Cha 41 .569 16.7 9.3 1.8 1.59

eW/G eW per36 rates tm mpg Eff% Sco Reb Ast e480
.135 .81 Stoudemire,Amare Phx 35 .580 26.0 6.7 1.7 1.86
.130 .52 Howard,Dwight Orl 27 .438 14.8 15.6 3.7 2.35
.129 .64 Pierce,Paul Bos 39 .576 22.5 6.0 3.1 1.61
.127 .89 Horford,Al Atl 36 .583 18.1 10.8 1.3 1.69
.126 .76 Millsap,Paul Uta 32 .612 18.8 11.9 1.7 1.90

.125 .75 Ginobili,Manu SA 33 .551 25.8 4.4 5.2 1.80
.124 .87 Smith,Josh Atl 36 .554 15.1 10.6 2.7 1.65
.121 .85 Johnson,Joe Atl 41 .476 19.7 5.2 4.1 1.42
.117 .58 Garnett,Kevin Bos 28 .600 20.2 10.2 2.6 2.01
.115 .69 Duncan,Tim SA 37 .498 19.9 9.9 2.4 1.48

.114 .57 Noah,Joakim Chi 38 .592 14.8 13.5 1.9 1.47
.113 .57 Jamison,Antawn Cle 35 .584 21.8 8.2 1.5 1.57
.113 .68 Nash,Steve Phx 33 .604 21.1 2.8 10.6 1.63
.111 .56 Rose,Derrick Chi 42 .493 20.9 3.1 4.8 1.27
.108 .65 Butler,Caron Dal 34 .524 24.9 6.7 1.4 1.53

eW/G eW per36 rates tm mpg Eff% Sco Reb Ast e480
.105 .63 Bynum,Andrew LA 30 .576 15.8 11.3 1.1 1.70
.101 .61 Billups,Chauncey Den 34 .587 22.1 2.8 5.8 1.43
.099 .39 Lewis,Rashard Orl 39 .685 20.4 6.3 2.0 1.23
.097 .68 Jennings,Brandon Mil 36 .492 19.7 3.3 3.3 1.32
.093 .56 Aldridge,Lamarcu Por 38 .496 16.6 6.4 1.8 1.17

.085 .51 Parker,Tony SA 31 .492 18.8 4.1 5.6 1.31
.085 .51 Hill,Grant Phx 27 .505 11.2 13.1 3.3 1.50
.079 .55 Salmons,John Mil 41 .484 15.4 3.5 3.2 .93
.075 .53 Crawford,Jamal Atl 30 .498 20.6 3.8 2.7 1.19
.071 .28 Barnes,Matt Orl 27 .548 13.7 8.6 3.2 1.25

.068 .41 Hill,George SA 35 .593 18.7 4.3 .6 .93
.067 .27 Thomas,Tyrus Cha 17 .648 20.9 12.7 .9 1.88
.066 .46 Ilyasova,Ersan Mil 22 .576 17.2 12.9 .6 1.42
.064 .32 Deng,Luol Chi 41 .505 14.3 4.8 1.0 .76
.062 .31 Williams,Mo Cle 37 .549 16.3 2.7 4.7 .81

.061 .37 Camby,Marcus Por 30 .427 5.7 14.0 2.6 1.00
.059 .30 O'Neal,Shaquille Cle 21 .535 18.3 11.8 3.0 1.39
.059 .41 Williams,Marvin Atl 31 .549 12.4 8.3 1.0 .92
.059 .23 Jackson,Stephen Cha 39 .450 16.3 5.4 3.2 .72
.058 .35 Miller,Andre Por 35 .500 15.5 3.7 5.4 .81

.058 .35 Kidd,Jason Dal 41 .462 7.4 6.3 5.9 .69
.058 .29 Davis,Glen Bos 22 .557 15.9 8.3 1.2 1.26

I figure eWins per game is the way to go.
During the season, there were just 10 in the league over .150, and 45 over .100 (1.5 per team) .
In the playoffs, there are 32 over .10, right at 2 per team.
LeBron's the only guy to go all season >.20 eW/G (.248)
Per minute, see eW/480, where 1.00 is 'average'.
_________________
`
36% of all statistics are wrong
Back to top
View user's profile Send private message Send e-mail
Mike G



Joined: 14 Jan 2005
Posts: 3553
Location: Hendersonville, NC

PostPosted: Tue May 04, 2010 4:57 pm Post subject: Reply with quote
Most Improved in the 2010 playoffs, relative to the season.
Playoff eWin/480 Min rates were boosted by .158 to create no net aggregate loss from regular season rates.
Code:
Wins above regular season Round 1 playoffs eWins/480 2010 Season
eW+ per36 rates tm Sco Reb Ast Rd1 RS Sco Reb Ast
.97 Richardson,Jason Phx 35.1 9.0 1.1 3.40 1.14 17.8 6.1 1.6
.55 Westbrook,Russel Okl 23.9 6.3 6.4 2.50 1.41 17.2 5.5 6.8
.54 Nelson,Jameer Orl 32.1 3.4 5.4 2.86 1.23 17.8 4.1 6.1
.39 Allen,Ray Bos 26.9 3.0 3.6 1.92 1.03 19.1 3.6 2.2
.31 Butler,Caron Dal 24.9 6.7 1.4 1.53 .96 15.9 5.9 1.5

.29 Gasol,Pau LAL 20.8 12.5 3.4 2.43 1.94 20.1 11.5 2.6
.29 Millsap,Paul Uta 18.8 11.9 1.7 1.90 1.34 15.8 9.5 1.6
.28 Williams,Deron Uta 25.5 2.7 8.7 2.11 1.72 20.4 4.3 8.1
.27 Hill,Grant Phx 11.2 13.1 3.3 1.50 .85 13.5 6.9 2.2
.26 Wade,Dwyane Mia 34.9 5.6 5.2 3.05 2.60 30.6 5.2 5.4

.25 James,Lebron Cle 34.6 8.7 6.7 3.47 3.05 32.3 7.5 6.4
.23 Rondo,Rajon Bos 14.3 6.1 9.0 1.78 1.41 14.9 4.9 7.8
.22 Nowitzki,Dirk Dal 32.5 8.0 2.7 2.38 2.09 26.1 7.7 2.0
.18 Horford,Al Atl 18.1 10.8 1.3 1.69 1.51 16.2 11.0 1.8
.18 Williams,Marvin Atl 12.4 8.3 1.0 .92 .69 12.3 6.6 1.0

.16 Davis,Glen Bos 15.9 8.3 1.2 1.26 .71 12.7 8.5 1.1
.15 Jennings,Brandon Mil 19.7 3.3 3.3 1.32 1.19 17.6 4.1 5.2
.14 Mbah a Moute,Luc Mil 13.8 8.5 .9 .79 .56 9.2 8.2 1.3
.14 Lewis,Rashard Orl 20.4 6.3 2.0 1.23 .95 17.0 5.3 1.4
.13 Pietrus,Mickael Orl 21.5 2.7 1.6 1.35 .66 14.4 4.9 .9


14 of 20 top improvers from teams which advance.
But I don't think any series was an upset.

Aggregate over/under for teams, sorted by conference and series.
West teams did worse than East teams on average; presumably due to greater competition.
Code:
West Conf. East Conf.
tm net tm net
Dal -.04 Atl .49
SAS -.34 Mil .26

Den -.17 Bos .50
Uta .02 Mia -.63

LAL .11 Cha -.47
Okl -.18 Orl .52

Phx 1.03 Chi -.34
Por -.81 Cle .11

_________________
`
36% of all statistics are wrong
Back to top
View user's profile Send private message Send e-mail
Mike G



Joined: 14 Jan 2005
Posts: 3553
Location: Hendersonville, NC

PostPosted: Sun May 09, 2010 5:44 pm Post subject: Reply with quote
After 3 games for everyone in round 2, most productive in those 3 games:
Code:
eW per36 rates tm mpg Eff% Sco Reb Ast Stl TO Blk e480
.89 James,Lebron Cle 41 .618 34.0 7.6 5.0 2.1 2.4 1.8 3.47
.59 Howard,Dwight Orl 35 .710 33.3 16.5 1.9 .3 3.5 2.4 2.66
.51 Duncan,Tim SA 38 .557 19.1 11.1 1.9 1.0 1.3 2.6 2.18
.51 Nash,Steve Phx 35 .630 27.2 5.4 6.9 .3 4.8 .0 2.32
.46 Richardson,Jason Phx 34 .638 27.7 5.6 1.0 1.1 1.8 .0 2.20

.45 Rondo,Rajon Bos 41 .630 19.5 5.1 9.5 .9 3.6 .0 1.77
.43 Bryant,Kobe LA 41 .615 28.1 3.6 4.6 .6 3.8 .9 1.70
.42 Gasol,Pau LA 40 .642 18.3 13.6 2.2 .6 3.0 2.7 1.68
.34 Ginobili,Manu SA 38 .613 21.0 2.9 5.0 2.2 3.5 .3 1.43
.34 Nelson,Jameer Orl 30 .654 27.9 3.0 6.7 .4 1.6 .0 1.80

.32 Lewis,Rashard Orl 37 .671 21.0 5.6 3.3 .7 1.0 .7 1.40
.32 Carter,Vince Orl 32 .593 24.6 6.6 3.4 .4 1.2 .4 1.62
.32 Garnett,Kevin Bos 35 .505 17.9 9.5 2.0 .7 1.8 1.1 1.47
.32 Millsap,Paul Uta 36 .561 17.0 7.8 2.7 1.3 1.3 .7 1.42
.31 Boozer,Carlos Uta 39 .460 14.5 11.9 2.5 .3 1.9 .3 1.29

eW per36 rates tm mpg Eff% Sco Reb Ast Stl TO Blk e480
.31 Williams,Deron Uta 38 .568 19.8 2.2 6.6 .3 2.5 .3 1.29
.30 Stoudemire,Amare Phx 37 .505 16.3 11.6 1.2 .0 2.6 1.3 1.27
.26 Jamison,Antawn Cle 34 .531 15.5 11.2 .7 .7 1.8 1.1 1.22
.24 Odom,Lamar LA 29 .571 11.1 14.9 2.1 .0 1.3 2.1 1.35
.23 Bynum,Andrew LA 25 .610 12.0 14.3 1.6 1.0 1.0 2.5 1.52

.23 Horford,Al Atl 32 .520 12.8 9.6 2.1 .4 .4 1.5 1.14
.23 Hill,Grant Phx 33 .614 17.8 6.2 2.0 1.1 2.2 .0 1.10
.22 Frye,Channing Phx 29 .789 15.9 6.2 1.8 .8 .4 1.6 1.20
.22 Dragic,Goran Phx 14 .566 24.9 6.2 3.6 1.6 1.6 .8 2.53
.21 Williams,Mo Cle 36 .486 11.8 3.9 6.3 .7 .7 .0 .93

_________________
`
36% of all statistics are wrong
Back to top
View user's profile Send private message Send e-mail
Mike G



Joined: 14 Jan 2005
Posts: 3553
Location: Hendersonville, NC

PostPosted: Tue May 11, 2010 8:47 am Post subject: Reply with quote
Round 2 leaders after 4 games each.
Code:
eW per36 rates tm mpg Eff% Sco Reb Ast Stl TO Blk e480
1.17 James,Lebron Cle 42 .583 31.5 7.7 5.5 2.0 3.3 1.6 3.39
.90 Rondo,Rajon Bos 42 .580 21.5 8.2 9.8 1.1 3.5 .0 2.55
.87 Gasol,Pau LA 40 .662 25.1 13.9 2.2 .5 2.5 2.5 2.61
.75 Bryant,Kobe LA 41 .598 32.1 3.5 4.4 .7 3.6 .7 2.19
.70 Howard,Dwight Orl 37 .724 28.0 14.5 1.6 .5 4.2 2.7 2.27

.68 Nash,Steve Phx 36 .633 25.8 5.0 7.2 .5 4.5 .0 2.28
.56 Duncan,Tim SA 37 .540 17.4 10.7 1.9 1.2 1.7 2.7 1.82
.51 Stoudemire,Amare Phx 38 .550 19.6 9.8 1.1 .5 1.9 1.2 1.62
.50 Nelson,Jameer Orl 30 .645 27.3 3.4 7.9 .9 1.8 .0 1.99
.48 Garnett,Kevin Bos 34 .533 19.1 9.2 2.1 .8 1.3 .8 1.70

eW per36 rates tm mpg Eff% Sco Reb Ast Stl TO Blk e480
.48 Lewis,Rashard Orl 36 .701 22.4 6.3 3.9 .8 1.0 .8 1.63
.46 Carter,Vince Orl 34 .629 24.7 6.1 3.2 .3 .8 .5 1.64
.45 Ginobili,Manu SA 38 .571 17.9 3.7 5.4 2.8 3.3 .2 1.40
.45 Richardson,Jason Phx 34 .600 22.1 6.4 1.0 .8 1.9 .3 1.57
.44 Williams,Deron Uta 40 .537 19.9 2.5 6.5 .7 2.8 .7 1.34

.42 Millsap,Paul Uta 33 .576 19.5 7.6 2.3 1.3 1.3 .8 1.54
.33 Jamison,Antawn Cle 35 .533 14.5 9.8 1.3 .5 1.3 1.1 1.15
.31 Smith,Josh Atl 35 .463 14.9 9.0 1.3 1.8 1.8 .5 1.08
.30 Bynum,Andrew LA 25 .564 10.8 13.4 1.6 .7 .7 2.6 1.44
.28 Odom,Lamar LA 28 .603 11.6 12.6 1.8 .0 .9 1.8 1.20
.28 Boozer,Carlos Uta 39 .453 12.1 12.3 1.9 .2 2.4 .2 .86

_________________
`
36% of all statistics are wrong
Back to top
View user's profile Send private message Send e-mail
Mike G



Joined: 14 Jan 2005
Posts: 3553
Location: Hendersonville, NC

PostPosted: Wed May 12, 2010 4:45 pm Post subject: Reply with quote
EDIT :
Final round-2 leaders
Code:
eW/G per36 rates tm mpg Eff% Sco Reb Ast Stl TO Blk e480
.232 James,Lebron Cle 42 .540 24.4 9.0 5.8 1.9 3.9 1.2 2.63
.216 Gasol,Pau LA 40 .662 25.1 13.9 2.2 .5 2.5 2.5 2.61
.208 Rondo,Rajon Bos 42 .579 20.9 6.3 9.4 1.6 3.2 .0 2.39
.187 Bryant,Kobe LA 41 .598 32.1 3.5 4.4 .7 3.6 .7 2.19
.176 Howard,Dwight Orl 37 .724 28.0 14.5 1.6 .5 4.2 2.7 2.27

.170 Nash,Steve Phx 36 .633 25.8 5.0 7.2 .5 4.5 .0 2.28
.143 Garnett,Kevin Bos 34 .551 21.2 9.7 2.1 .5 1.4 .9 2.00
.141 Duncan,Tim SA 37 .540 17.4 10.7 1.9 1.2 1.7 2.7 1.82
.127 Stoudemire,Amare Phx 38 .550 19.6 9.8 1.1 .5 1.9 1.2 1.62
.125 Nelson,Jameer Orl 30 .645 27.3 3.4 7.9 .9 1.8 .0 1.99

.120 Lewis,Rashard Orl 36 .701 22.4 6.3 3.9 .8 1.0 .8 1.63
.115 Carter,Vince Orl 34 .629 24.7 6.1 3.2 .3 .8 .5 1.64
.111 Ginobili,Manu SA 38 .571 17.9 3.7 5.4 2.8 3.3 .2 1.40
.111 Richardson,Jason Phx 34 .600 22.1 6.4 1.0 .8 1.9 .3 1.57
.111 Williams,Deron Uta 40 .537 19.9 2.5 6.5 .7 2.8 .7 1.34

.106 Millsap,Paul Uta 33 .576 19.5 7.6 2.3 1.3 1.3 .8 1.54
.078 Smith,Josh Atl 35 .463 14.9 9.0 1.3 1.8 1.8 .5 1.08
.075 O'Neal,Shaquille Cle 24 .551 21.5 8.5 1.4 .3 2.5 1.5 1.53
.074 Bynum,Andrew LA 25 .564 10.8 13.4 1.6 .7 .7 2.6 1.44
.070 Odom,Lamar LA 28 .603 11.6 12.6 1.8 .0 .9 1.8 1.20

_________________
`
36% of all statistics are wrong
Back to top
View user's profile Send private message Send e-mail
Mike G



Joined: 14 Jan 2005
Posts: 3553
Location: Hendersonville, NC

PostPosted: Sun Jun 06, 2010 7:24 am Post subject: Reply with quote
eWins leaders thru the conference finals:
Code:
eWins per36 rates tm G mpg Eff% Sco Reb Ast Stl TO Blk e480
2.90 Bryant,Kobe LA 16 40 .581 29.0 5.1 4.7 .9 3.0 .7 2.19
2.85 James,Lebron Cle 11 42 .600 29.0 8.9 6.2 1.5 3.3 1.6 2.99
2.58 Rondo,Rajon Bos 17 41 .522 16.3 5.3 8.4 1.8 2.8 .1 1.76
2.57 Gasol,Pau LA 16 39 .608 20.4 10.8 2.7 .2 1.8 1.8 1.99
2.27 Howard,Dwight Orl 14 36 .605 22.7 13.1 1.6 .8 3.8 3.6 2.20

2.14 Richardson,Jason Phx 16 33 .634 24.1 6.7 1.0 1.2 .7 .3 1.93
2.13 Pierce,Paul Bos 17 38 .585 21.2 6.6 3.2 1.1 2.6 .4 1.56
2.06 Nash,Steve Phx 16 34 .621 21.8 3.9 9.7 .3 4.1 .1 1.83
1.92 Stoudemire,Amare Phx 16 37 .573 23.0 7.3 1.0 .7 2.5 1.5 1.58
1.90 Garnett,Kevin Bos 17 32 .534 17.1 9.9 2.4 1.1 1.3 .7 1.68

1.86 Nelson,Jameer Orl 14 34 .608 25.5 4.4 5.7 1.1 2.7 .0 1.87
1.54 Allen,Ray Bos 17 38 .603 18.5 3.8 2.8 1.0 1.3 .2 1.14
1.49 Williams,Deron Uta 10 40 .600 23.3 2.7 7.8 .9 2.8 .4 1.80
1.32 Wade,Dwyane Mia 5 42 .638 34.9 5.6 5.2 1.4 4.6 1.4 3.05
1.25 Duncan,Tim SA 10 37 .513 18.9 10.2 2.2 .8 2.4 1.7 1.62

eWins per36 rates tm G mpg Eff% Sco Reb Ast Stl TO Blk e480
1.24 Boozer,Carlos Uta 10 40 .542 16.6 12.9 2.2 .4 2.2 .6 1.48
1.19 Ginobili,Manu SA 10 35 .558 22.4 4.1 5.3 2.7 3.2 .2 1.63
1.19 Bynum,Andrew LA 16 24 .584 14.0 12.3 1.0 .5 1.4 2.6 1.47
1.18 Millsap,Paul Uta 10 33 .597 19.1 10.1 1.9 1.2 1.4 1.5 1.75
1.18 Smith,Josh Atl 11 36 .522 15.0 10.1 2.2 1.2 1.6 1.8 1.45

1.16 Horford,Al Atl 11 35 .566 16.2 10.2 1.5 .7 1.4 1.8 1.43
1.15 Nowitzki,Dirk Dal 5 39 .630 32.5 8.0 2.7 .8 1.6 .6 2.38
1.12 Odom,Lamar LA 16 30 .514 11.4 11.9 2.2 .8 1.3 1.2 1.13
1.11 Westbrook,Russel Okl 6 36 .559 23.9 6.3 6.4 1.7 2.4 .2 2.50
1.02 Anthony,Carmelo Den 6 42 .553 25.7 8.4 2.3 1.7 3.0 .4 1.96

1.01 Lewis,Rashard Orl 14 37 .634 15.4 6.5 2.5 1.1 1.8 .7 .95
.99 Carter,Vince Orl 14 34 .518 17.9 5.2 2.7 .9 1.7 .2 .99
.97 Johnson,Joe Atl 11 40 .455 16.4 5.0 3.7 .8 2.0 .2 1.06
.90 Durant,Kevin Okl 6 39 .487 24.6 7.4 2.3 .5 3.5 1.3 1.86
.88 Hill,Grant Phx 16 28 .546 12.3 8.5 2.6 1.1 1.5 .8 .93

eWins/480 minutes less than 1.0 is subpar. Over 2.0 could be called superstar.
_________________
`
36% of all statistics are wrong
Back to top
View user's profile Send private message Send e-mail
Mike G



Joined: 14 Jan 2005
Posts: 3553
Location: Hendersonville, NC

PostPosted: Tue Jun 22, 2010 7:39 am Post subject: Reply with quote
Total postseason eWins leaders for 2010
Code:
eWins per36 rates tm G Min Eff% Sco Reb Ast Stl TO Blk e480
4.49 Bryant,Kobe LA 23 40 .558 29.1 6.0 4.3 1.2 3.1 .7 2.34
3.94 Gasol,Pau LA 23 40 .584 19.8 11.0 2.9 .4 1.8 1.9 2.07
3.32 Rondo,Rajon Bos 24 41 .495 15.4 5.5 7.9 1.7 2.7 .1 1.64
2.85 James,Lebron Cle 11 42 .590 29.0 8.9 6.2 1.5 3.3 1.6 2.99
2.79 Pierce,Paul Bos 24 39 .554 20.4 6.2 3.0 .9 2.5 .5 1.43

2.75 Garnett,Kevin Bos 23 33 .524 17.9 9.0 2.6 1.2 1.5 1.0 1.72
2.27 Howard,Dwight Orl 14 36 .580 22.7 13.1 1.6 .8 3.8 3.6 2.20
2.14 Richardson,Jason Phx 16 33 .627 24.1 6.7 1.0 1.2 .7 .3 1.93
2.06 Nash,Steve Phx 16 34 .621 21.8 3.9 9.7 .3 4.1 .1 1.83
1.92 Stoudemire,Amare Phx 16 37 .573 23.0 7.3 1.0 .7 2.5 1.5 1.58

1.86 Nelson,Jameer Orl 14 34 .602 25.5 4.4 5.7 1.1 2.7 .0 1.87
1.70 Allen,Ray Bos 24 39 .563 17.1 3.5 2.4 .9 1.4 .1 .88
1.49 Williams,Deron Uta 10 40 .594 23.3 2.7 7.8 .9 2.8 .4 1.80
1.44 Bynum,Andrew LA 23 24 .559 13.2 11.2 .7 .4 1.4 2.4 1.23
1.40 Odom,Lamar LA 23 29 .507 11.1 11.2 2.1 .8 1.5 1.1 1.01

eWins per36 rates tm G Min Eff% Sco Reb Ast Stl TO Blk e480
1.32 Wade,Dwyane Mia 5 42 .638 34.9 5.6 5.2 1.4 4.6 1.4 3.05
1.26 Duncan,Tim SA 10 37 .515 18.9 10.2 2.2 .8 2.4 1.7 1.63
1.24 Boozer,Carlos Uta 10 40 .531 16.6 12.9 2.2 .4 2.2 .6 1.48
1.23 Nowitzki,Dirk Dal 6 39 .630 34.0 8.0 2.7 .8 1.6 .6 2.54
1.21 Ginobili,Manu SA 10 35 .559 22.7 4.1 5.3 2.7 3.2 .2 1.65

1.18 Millsap,Paul Uta 10 33 .596 19.1 10.1 1.9 1.2 1.4 1.5 1.75
1.18 Smith,Josh Atl 11 36 .513 15.0 10.1 2.2 1.2 1.6 1.8 1.45
1.16 Horford,Al Atl 11 35 .561 16.2 10.2 1.5 .7 1.4 1.8 1.43
1.11 Westbrook,Russel Okl 6 36 .559 23.9 6.3 6.4 1.7 2.4 .2 2.50
1.02 Anthony,Carmelo Den 6 42 .553 25.7 8.4 2.3 1.7 3.0 .4 1.96

1.01 Lewis,Rashard Orl 14 37 .579 15.4 6.5 2.5 1.1 1.8 .7 .95
.99 Carter,Vince Orl 14 34 .497 17.9 5.2 2.7 .9 1.7 .2 .99
.97 Johnson,Joe Atl 11 40 .448 16.4 5.0 3.7 .8 2.0 .2 1.06
.90 Durant,Kevin Okl 6 39 .487 24.6 7.4 2.3 .5 3.5 1.3 1.86
.88 Hill,Grant Phx 16 28 .542 12.3 8.5 2.6 1.1 1.5 .8 .93

eWins per36 rates tm G Min Eff% Sco Reb Ast Stl TO Blk e480
.85 Davis,Glen Bos 24 20 .535 13.4 8.4 .7 1.3 1.5 .6 .85
.84 Jamison,Antawn Cle 11 34 .533 16.3 8.6 1.3 .7 1.8 1.1 1.08
.75 O'Neal,Shaquille Cle 11 22 .545 20.1 9.9 2.1 .3 3.6 1.9 1.47
.71 Parker,Tony SA 10 34 .494 17.4 4.1 4.8 .6 2.3 .0 1.02
.69 Crawford,Jamal Atl 11 32 .489 18.6 3.3 2.5 .9 2.0 .1 .94

.68 Jennings,Brandon Mil 7 36 .492 19.7 3.3 3.3 1.2 1.2 .6 1.32
.66 Butler,Caron Dal 6 34 .524 25.1 6.7 1.4 1.7 2.8 .9 1.55
.61 Billups,Chauncey Den 6 34 .587 22.1 2.8 5.8 1.1 3.2 .5 1.43
.61 Artest,Ron LA 23 37 .473 10.0 4.3 1.9 1.5 1.1 .5 .35
.60 Dudley,Jared Phx 16 24 .610 11.7 6.1 2.3 1.7 .8 .5 .77
.60 Williams,Mo Cle 11 37 .527 14.2 3.3 4.9 .5 2.0 .2 .70

_________________
`
36% of all statistics are wrong
Back to top
View user's profile Send private message Send e-mail
Manchvegasbob



Joined: 03 Aug 2008
Posts: 52


PostPosted: Wed Jun 23, 2010 7:05 am Post subject: Reply with quote
Thanks Mike - you've saved me some trouble regarding research on folks - your stat summaries are always quotable.

Ilardi



Joined: 15 May 2008
Posts: 265
Location: Lawrence, KS

PostPosted: Mon May 19, 2008 8:51 am Post subject: Additional Homecourt Advantage in Playoffs (Points)? Reply with quote
As I understand it, most predictive models (including those used by Vegas oddsmakers) yield a homecourt advantage of roughly 3.5 points during the NBA regular season; Sagarin's model, for example, put the advantage at roughly 3.6 points at end of the 07-08 season. But it looks to me like the advantage may be larger during the playoffs (attributable, perhaps, to the effect of increased crowd noise on refs and/or players?), but I haven't taken the time to try modeling the increase. (For what it's worth, I see that Sagarin's model today - covering both regular season and playoffs - had bumped up the homecourt advantage to over 4.0 points per game from the regular season value of 3.6, suggesting that the playoff-based advantage must be larger than that of the regular season, at least this year.)

Has anyone out there ever modeled this effect? If so, how large is it? Also, is this year's apparent bump in homecourt advantage thus far an aberration, or part of a regular pattern observable each year?

Any leads on this topic would be appreciated.
Back to top
View user's profile Send private message
Serhat Ugur (hoopseng)



Joined: 13 Oct 2006
Posts: 209
Location: Basketball Research

PostPosted: Mon May 19, 2008 10:38 am Post subject: Reply with quote
According to my research, the referees are not a factor.

http://www.nbastuffer.com/referee_stats

51.5% of fouls are being called against road teams while home teams have slight advantage of getting to the foul line more
_________________
http://www.nbastuffer.com
Back to top
View user's profile Send private message Visit poster's website
Ilardi



Joined: 15 May 2008
Posts: 265
Location: Lawrence, KS

PostPosted: Mon May 19, 2008 12:43 pm Post subject: Reply with quote
There's a pretty extensive research literature out there documenting the ability of home crowds to induce referee bias across a number of sports, including basketball (see sample citations below). The stats you provide below are congruent with that biasing effect, despite the apparent small magnitude of the effect size.

However, it's important to bear in mind that the referee bias could still be rather large, despite the modest observed disparity in fouls called (i.e., 51.5% called against road teams, 48.5% called against home teams). For example, if refs really are much more likely to call fouls against the visiting team in the playoffs, players on both may simply *adjust* their play accordingly - with visiting players being more timid and home players more aggressive - such that the total number of fouls called looks fairly even on both sides.

On large, ambiguous, "judgment-call" decisions in soccer (e.g., awarding penalty kicks), there is now pretty uncontrovertible evidence that refs are influenced by the home crowd. It would be surprising if this effect did not also exert an influence in basketball, where the venue is even more intimate and the fans much closer.



Courneya, K. S., & Carron, A. V. (1992). The home advantage in sport competitions: A literature review. Journal of Sport & Exercise Psychology, 14, 13-27.

Greer, D. L. (1983). Spectator booing and the home advantage: A study of social influence in the basketball arena. Social Psychology Quarterly, 46, 252-261.

Lehman, D. R., & Reifman, A. (1987). Spectator influence on basketball officiating. The Journal of Social Psychology, 127, 673-675.

Madrigal, R., & James, J. (1999). Team quality and the home advantage. Journal of Sport Behavior, 22, 381-399.

Nevill, A. M., Balmer, N. J., & Williams, A. M. (2002). The influence of crowd noise and experience upon refereeing decisions in football. Psychology of Sport and Exercise, 3, 261-272.

Nevill, A. M. & Holder, R. L. (1999). Home advantage in sport: An overview of studies on the advantage of playing at home. Sports Medicine, 28, 221-236.

Silva, J. M., & Andrew, J. A. (1987). An analysis of game location and basketball performance in the Atlantic Coast Conference. International Journal of Sport Psychology, 18, 188-204.

Last edited by Ilardi on Tue May 20, 2008 10:52 am; edited 1 time in total
Back to top
View user's profile Send private message
Charles



Joined: 16 May 2005
Posts: 134


PostPosted: Tue May 20, 2008 8:18 am Post subject: Reply with quote
I have some numbers from 1988 to 2007. During the regular season the home team won by a margin of about 3.6.

Code:
Regular season
Expected Actual Difference
Home 99.36 101.17 +1.81
Away 99.36 97.56 -1.81
Margin 0.00 3.61 3.61


In the playoffs, the home teams have a slight edge, based on regular season scoring differentials (although the difference of just .16, is somewhat smaller than I would have expected.) The actual margin for home teams in the playoffs was 4.2.

Code:
Playoffs
Expected Actual Difference
Home 95.50 95.55 +0.06
Away 95.34 91.20 -4.14
Margin 0.16 4.36 4.20


Incidentally, once you account for the fact that the pace of playoff games dropped from a regular season average pace of 95.8 to just 91.0, the scoring efficiency per 100 possessions was almost identical in the post season.
Back to top
View user's profile Send private message
Charles



Joined: 16 May 2005
Posts: 134


PostPosted: Tue May 20, 2008 9:01 am Post subject: Reply with quote
Here are the changes in some team stats from the regular season to the playoffs. Most are shown as home team minus away team per 100 possessions.

During the regular season home teams drew .8 less personal fouls and took 1.3 more free throws per 100 possessions than away teams. In the playoffs those advantages increased to 1.4 less fouls against and 2.2 more free throws. It's impossible to know whether this small change is due to difference in refereeing or whether teams might play more aggressively on the road during the playoffs.

Code:
Home team - Away team difference (per 100 possessions)

PF FTA
Regular season -0.8 1.3
Playoff -1.4 2.2



The home team advantage in offensive rebounding decreased somewhat in the playoffs, but their advantage in turnover rate increased some. Again, both of these might indicate somewhat more aggression on the part of away teams in the playoffs, relative to the regular season.

Code:
OREB AST TO ST BLK
Regular season 0.6 2.2 -0.4 0.2 0.8
Playoff 0.3 2.2 -0.9 0.2 0.8


-- Field Goal Percentage --
Home Away Diff
Regular season .464 .451 .013
Playoff .447 .431 .017



You would expect any differences due to environmental factors like crowd noise to present themselves at the free throw line, but during the regular season teams shot free throws slightly better at home, whereas, in the playoffs, teams actually shot slightly better on the road. Of course, the differences are very small and it is possible that the distribution of attempts by specific players home vs away might vary slightly from the regular season to the playoffs.

Code:
-- Free Throw Percentage --
Home Away Diff
Regular season .751 .748 0.003
Playoff .743 .744 -0.001



It appears that the slightly increase in point differential enjoyed by home teams during the playoffs is attributable to a few extra free throws and slightly fewer turnovers.
Back to top
View user's profile Send private message
Serhat Ugur (hoopseng)



Joined: 13 Oct 2006
Posts: 209
Location: Basketball Research

PostPosted: Tue May 20, 2008 10:14 am Post subject: Reply with quote
Charles,

Do those numbers belong to 2007-2008 regular season and 2008 playoffs?
_________________
http://www.nbastuffer.com
Back to top
View user's profile Send private message Visit poster's website
Ryan J. Parker



Joined: 23 Mar 2007
Posts: 711
Location: Raleigh, NC

PostPosted: Tue May 20, 2008 10:27 am Post subject: Reply with quote
Also, are these statistically significant differences?
Back to top
View user's profile Send private message Visit poster's website
Charles



Joined: 16 May 2005
Posts: 134


PostPosted: Wed May 21, 2008 1:01 pm Post subject: Reply with quote
The sample is 1997-98 through 2006-07. It does not include this season. The samples are very large, so most are significant (<.001). Even the tiny free throw differential is significant. I was surprised by a few of the numbers. I would be interested in what other people have found in regard to how the game changes in the playoffs. Also, I could run this sample some other way if you have any specific ideas.
Back to top
View user's profile Send private message
Serhat Ugur (hoopseng)



Joined: 13 Oct 2006
Posts: 209
Location: Basketball Research

PostPosted: Wed May 21, 2008 1:17 pm Post subject: Reply with quote
This year's playoffs and regular season comparison.

http://www.nbastuffer.com/What_Changed_In_Playoffs.html
_________________
http://www.nbastuffer.com
Back to top
View user's profile Send private message Visit poster's website
Ryan J. Parker



Joined: 23 Mar 2007
Posts: 711
Location: Raleigh, NC

PostPosted: Wed May 21, 2008 1:18 pm Post subject: Reply with quote
What is the advantage in terms of winning %? By that I mean how often do home teams win in the regular season compared to the post season?

Can any of this be attributed to the notion that better teams have home court and thus we are more likely to see better teams play on a home court? I don't have series outcomes handy, so that might be something to look at.
Back to top
View user's profile Send private message Visit poster's website
Mike G



Joined: 14 Jan 2005
Posts: 3615
Location: Hendersonville, NC

PostPosted: Wed May 21, 2008 3:34 pm Post subject: Reply with quote
Ryan J. Parker wrote:

Can any of this be attributed to the notion that better teams have home court and thus we are more likely to see better teams play on a home court?

Heck, yes.
_________________
`
36% of all statistics are wrong
Back to top
View user's profile Send private message Send e-mail
Charles



Joined: 16 May 2005
Posts: 134


PostPosted: Wed May 21, 2008 3:55 pm Post subject: Reply with quote
Ryan J. Parker wrote:
What is the advantage in terms of winning %? By that I mean how often do home teams win in the regular season compared to the post season?

Can any of this be attributed to the notion that better teams have home court and thus we are more likely to see better teams play on a home court? I don't have series outcomes handy, so that might be something to look at.


I just estimated the home team's expected win percentage by averaging the home team's win percentage and the away team's loss percentage. I'm sure there are better ways, but hey...
Code:
--- Home Team Win Percentage ---
Expected Actual Difference
Regular season .500 .611 +.111
Playoffs .506 .645 +.139

Again, I am surprised that playoff home team's regular season records are, on average, just slightly better than away teams. They certainly win a higher percentage of their playoff games than their regular season records would suggest. What would be a better way of estimating the home team's chance of winning based on the two team's regular season records?

Mike G



Joined: 14 Jan 2005
Posts: 3630
Location: Hendersonville, NC

PostPosted: Thu Aug 03, 2006 9:10 am Post subject: Predicting Playoffs from past player performance. Reply with quote
Playoffs can be standardized just as regular seasons can be, at least the way I do them. Whether or not an assist is worth more than a point, we can compare the Season Kobe to the Playoff Kobe, with pretty fair accuracy.

Intuitively, it seems players tend to establish themselves as 'playoff guys' -- or not -- fairly early in their careers. Anecdotally and analytically, Horry has always been good, and Cliff Robinson has always been bad. I don't know if GM's or coaches tend to stock their rosters with the former; but I would tend to, myself.

After standardization, it seems this last postseason saw players about 93.5% as productive (per-minute) as they had been in the season. If players are going to 'regress to the mean', I guess they'll regress to .935.

I got a pretty good fit for this year's results by splitting a player's historic rate (which I call PO/RS) with the .935 figure. The basic form for expected PO/RS is :

exp = .20*pre + .80*(.935)

where 'pre' is the players Previous PO/RS, for years past. This is only given 20% of the weight. The other 80% is the constant (.8*.935=) .748

So (if I had done this before the '06 playoffs), I wouldn't expect DWade to reproduce his historic 1.05 PO/RS, but :

.2*1.05 + .748 = .958

Assigning these expectations to all players with a history of at least 65 playoff minutes, I noticed another trend: Older players tend to drop off more, in the playoffs. I've actually noticed this before, though I've not quantified it. Before his last 2 playoffs, Shaq had been a monster, after his first showing. Now he's headed for a very ordinary career PO/RS.

This is my first go-round, so I have arbitrarily guessed that age 32 is when players might be expected to show the first signs of 'playoff age fatigue'. So my full formula is now:

exp = .2*pre + .748 - (age-31)/120

All players age 31 or less were assigned age=31; so their adjustment is zero. Dividing by 120 made everyone's (expected minus actual) sum to zero.

Coming in, the Spurs led the league in individual playoff history (23,819 minutes, among guys who had at least 65 prior minutes AND at least 20 minutes in '06).

The Spurs netted -94 'production units' in 13 games. This is equivalent to about -62 points, or 4.5 per game. (They dropped 2 against the Kings, and went a mediocre 7-6 overall). Leading the dropoff was Tony Parker, whose .90 history was bankrupted (net -71) by his .78 showing. Ginobili came in as a 1.10 superplayoff guy, and came out at .88, for a -39. Duncan upstaged his own 1.02 previous with a 1.06 performance; but it (+60) wasn't enough.

Some part-rosters listed below had very few players with significant playoff experience; so any predictability is minor, at best. I'm not showing their prior PO/RS; it's just included in their (xP) 'expected playoff production in given mintues' column. Make of it what you will.

Oh yeah, the first 2 columns are RS and PO 'rates', by my system. Then expected and actual POProduction, and difference. Players in descending order of previous playoff minutes
Code:
Tm RS PO xP POP net player

Chi 27.7 31.0 178 201 23 Kirk Hinrich
Chi 26.2 24.2 166 165 -1 Ben Gordon
Chi 22.2 12.0 65 35 -30 Tyson Chandler
Chi 12.0 22.4 8 17 9 Eric Piatkowski

Chi 6 games 417 418 1

note: Gordon's (near-zero) net (-1) means I would predict his numbers (26.2) to fall off (24.2) to 92% of his RS rate; which they did. Meanwhile, Hinrich outdid himself, while Chandler fainted.


Cle 16.0 15.8 170 180 9 Eric Snow
Cle 15.1 9.6 73 49 -25 Damon Jones
Cle 25.7 21.0 224 196 -29 Larry Hughes
Cle 22.3 22.4 196 216 20 Donyell Marshall
Cle 28.9 28.4 216 221 5 Drew Gooden
Cle 34.6 27.0 315 265 -50 Zydrunas Ilgauskas

Cle 13 G 1194 1126 -68


Dal 43.7 40.9 1108 1117 8 Dirk Nowitzki
Dal 24.8 21.8 453 430 -23 Jerry Stackhouse
Dal 22.0 15.3 96 74 -22 Keith Vanhorn
Dal 29.1 27.3 649 641 -8 Jason Terry
Dal 31.2 28.9 682 662 -20 Josh Howard
Dal 15.5 15.8 17 21 3 Darrel Armstrong
Dal 24.3 22.4 286 283 -4 Erick Dampier

Dal 23 G 3291 3227 -64


Den 27.5 27.7 25 27 2 Kenyon Martin
Den 34.7 31.9 159 156 -3 Marcus Camby
Den 26.9 29.7 128 151 23 Andre Miller
Den 33.1 26.4 160 141 -19 Carmelo Anthony
Den 23.7 17.2 35 28 -7 Ruben Patterson

Den 5 G 507 502 -5


Det 30.0 26.4 492 461 -30 Rasheed Wallace
Det 27.3 23.8 461 424 -37 Ben Wallace
Det 28.7 26.9 520 515 -5 Richard Hamilton
Det 37.9 33.2 704 650 -54 Chauncey Billups
Det 23.6 26.7 470 553 83 Tayshaun Prince
Det 19.9 24.6 105 149 44 Lindsey Hunter
Det 26.3 22.0 97 85 -13 Tony Delk
Det 24.0 27.8 234 286 52 Antonio Mcdyess

Det 18 G 3082 3123 40


Ind 30.0 29.5 39 42 3 Peja Stojakovic
Ind 36.0 32.9 198 197 -1 Jermaine O'Neal
Ind 24.9 19.6 147 124 -23 Stephen Jackson
Ind 24.2 17.3 109 84 -25 Austin Croshere
Ind 24.9 31.6 156 211 56 Anthony Johnson
Ind 24.5 18.0 4 3 -1 Jamaal Tinsley
Ind 24.1 24.2 51 54 3 Jeff Foster

Ind 6 G 704 716 12


LAC 31.7 31.9 314 358 43 Sam Cassell
LAC 22.5 18.4 269 241 -28 Cuttino Mobley
LAC 23.3 25.3 143 173 30 Vladimir Radmanovic
LAC 17.3 4.5 9 3 -7 Zeljko Rebraca

LAC 12 G 736 775 39


LAL 43.4 30.1 355 264 -91 Kobe Bryant
LAL 10.6 13.8 6 8 2 Jim Jackson
LAL 28.7 28.3 231 247 16 Lamar Odom

LAL 7 G 592 519 -73


Mem 24.9 22.3 75 74 -1 Eddie Jones
Mem 27.3 10.0 71 28 -43 Bobby Jackson
Mem 26.3 19.4 71 56 -16 Chucky Atkins
Mem 29.2 17.6 77 52 -25 Mike Miller
Mem 38.6 28.1 158 123 -35 Pau Gasol
Mem 21.9 21.6 48 52 3 Lorenzen Wright
Mem 23.8 14.9 74 53 -21 Shane Battier
Mem 17.6 17.4 9 11 1 Brian Cardinal

Mem 4 G 585 448 -136


Mia 39.9 35.6 774 750 -25 Shaquille O'Neal
Mia 18.0 16.4 245 256 11 Gary Payton
Mia 30.6 28.3 171 177 6 Alonzo Mourning
Mia 25.2 23.1 557 553 -4 Antoine Walker
Mia 42.2 44.2 1075 1176 101 Dwyane Wade
Mia 25.0 21.7 442 414 -28 Jason Williams
Mia 22.1 21.8 366 393 27 Udonis Haslem
Mia 11.5 8.0 26 20 -6 Shandon Anderson
Mia 17.6 20.5 273 344 71 James Posey

Mia 23 G 3931 4084 153


Mil 20.1 23.0 26 34 8 Toni Kukoc
Mil 23.9 22.3 63 65 2 Joe Smith
Mil 22.1 25.5 78 96 17 Jamaal Magloire
Mil 31.1 34.1 142 175 34 Michael Redd

Mil 5 G 309 370 61


NJ 16.9 13.0 80 72 -8 Clifford Robinson
NJ 33.0 30.8 379 384 5 Jason Kidd
NJ 29.4 30.5 329 371 42 Richard Jefferson
NJ 12.6 10.1 98 85 -13 Jason Collins
NJ 37.1 44.3 433 554 121 Vince Carter
NJ 19.3 18.3 97 100 4 Lamond Murray

NJ 11 G 1415 1566 151


Phe 38.3 32.1 787 710 -77 Steve Nash
Phe 35.6 32.5 768 766 -2 Shawn Marion
Phe 22.6 27.7 380 486 107 Tim Thomas

Phe 20 G 1935 1963 28



SA 23.2 20.6 133 127 -6 Robert Horry
SA 41.7 44.1 542 602 60 Tim Duncan
SA 36.8 28.8 450 379 -71 Tony Parker
SA 17.2 12.7 190 156 -34 Bruce Bowen
SA 22.0 19.3 231 221 -9 Michael Finley
SA 21.8 14.5 73 54 -20 Nick Vanexel
SA 37.2 32.6 426 387 -39 Manu Ginobili
SA 23.7 21.4 174 179 6 Brent Barry
SA 20.3 19.8 59 64 6 Rasho Nesterovic
SA 26.4 30.8 65 80 15 Nazr Mohammed

SA 13 G 2345 2251 -94


Sac 29.4 19.9 199 142 -57 Mike Bibby
Sac 26.2 25.7 137 142 5 Ron Artest
Sac 28.6 20.9 121 97 -24 Brad Miller
Sac 28.2 32.1 183 222 39 Bonzi Wells
Sac 24.8 16.4 96 68 -28 Kenny Thomas
Sac 11.8 19.9 3 6 3 Vitaly Potapenko

Sac 6 G 738 675 -63


Was 21.3 20.9 123 126 3 Antonio Daniels
Was 29.2 25.2 191 177 -14 Antawn Jamison
Was 36.8 38.9 267 308 41 Gilbert Arenas
Was 23.7 19.8 99 85 -14 Brendan Haywood

Was 6 G 680 696 16

_________________
`
36% of all statistics are wrong
Back to top
View user's profile Send private message Send e-mail
ziller



Joined: 30 Jun 2005
Posts: 126
Location: Sac Metro

PostPosted: Thu Aug 03, 2006 11:12 am Post subject: Reply with quote
I knew Mike Bibby had a bad series, but I had no idea he was that much of a disaster.
_________________
SactownRoyalty.com
tziller@gmail.com
Back to top
View user's profile Send private message Send e-mail Visit poster's website
Ben



Joined: 13 Jan 2005
Posts: 266
Location: Iowa City

PostPosted: Thu Aug 03, 2006 11:55 am Post subject: Re: Predicting Playoffs from past player performance. Reply with quote
Mike G wrote:


exp = .20*pre + .80*(.935)


This is interesting stuff. Where did you get the .2 from? A regression?

How did Lebron do? >.935?
Back to top
View user's profile Send private message
Mark



Joined: 20 Aug 2005
Posts: 807


PostPosted: Thu Aug 03, 2006 11:58 am Post subject: Reply with quote
Good stuff as usual. Have you looked for trends regarding reg. season to playoff change by leading stat (pts, assists, rebs) which changed more or less? Or position or bodytype for position? Any commonalities for the guys who exceed regular season in playoffs- athletic?, wiley/lots of counter moves? went to major college / got big game experience?, got being 'the man' experience? Hard to say but some of these guys might benefit from defensive efforts to shut down other teammates.
Back to top
View user's profile Send private message
Mike G



Joined: 14 Jan 2005
Posts: 3630
Location: Hendersonville, NC

PostPosted: Thu Aug 03, 2006 12:37 pm Post subject: Re: Predicting Playoffs from past player performance. Reply with quote
Ben wrote:
Mike G wrote:


exp = .20*pre + .80*(.935)


This is interesting stuff. Where did you get the .2 from? A regression?

How did Lebron do? >.935?


Yeah, it was problematic to weigh the relative significance of prior playoff minutes and this year's minutes. I just did regression on subsets like:
Code:
prev '06 SS fac
3000 300 12 .060
3000 100 14 .004

2000 600 11 .240
2000 400 21 .224
2000 200 26 .222
2000 100 29 .220

1000 700 11 .329
1000 500 18 .255
1000 300 32 .260
1000 100 52 .050


The 1st 2 columns are minutes-qualifiers, for career-previous and for this season. SS is the sample size for each subset. So, with 10 different subsets, Shaq gets counted 10 times, and Damon Jones is only counted once.

The additive mean of the 'factor' column is only .186
The multiplicative mean is only .121, thanks to the .004 up there.
The non-correlation with previous, for guys with >3000 minutes, is what alerted me to the need to reduce expectations for older players. I guess I could use .25 .

Lebron starts his playoff career with a .85 vs his '06 RS. Note, it came vs the Pistons.

Mike Bibby: needed 43 mpg to average 16.7 pts, eff% down .100 from RS. Sac gave up 107 ppg (.601 SA eff%). So his 'effective scoring' (per36) drops from 21.2 to 12.7; assists also down about 1/3.

Coming in, he was a career 1.02 PO/RS guy; and he tanked at .68. Hence the shortfall you see.
_________________
`
36% of all statistics are wrong
Back to top
View user's profile Send private message Send e-mail
Mike G



Joined: 14 Jan 2005
Posts: 3630
Location: Hendersonville, NC

PostPosted: Fri Aug 04, 2006 4:10 am Post subject: Reply with quote
Mark wrote:
Good stuff as usual. Have you looked for trends regarding reg. season to playoff change by leading stat (pts, assists, rebs) which changed more or less? Or position or bodytype for position? Any commonalities for the guys who exceed regular season in playoffs- athletic?, wiley/lots of counter moves? went to major college / got big game experience?, got being 'the man' experience? Hard to say but some of these guys might benefit from defensive efforts to shut down other teammates.


I don't try to quantify 'bodytype', 'athletic', 'wiley', 'major', 'big', etc. I'd assumed the NBA playoffs is the only relevant experience, but maybe NCAA tourney helps -- at least back when it was competitive.

Versatility can be quantified; that seems at least to buy minutes.

Players change series by series, for sure; due mostly to matchups, I assume. Several Spurs who rampaged vs Sac, disappeared vs Dallas. But I don't anticipate doing breakdowns like this, in any finite amount of time.

The later rounds have been shown to produce lower average PO/RS rates, of course. Rebound rates, by definition, will be harder to maintain against stronger rebounding teams. Assists have lately become universally rarer. Effective shooting usually drops. These specifics depend on the teams. So I might determine that, for example, Reb tend to be hit hardest; yet the next opponent's rebounding strength would be a stronger determiner.
_________________
`
36% of all statistics are wrong
Back to top
View user's profile Send private message Send e-mail
Mike G



Joined: 14 Jan 2005
Posts: 3630
Location: Hendersonville, NC

PostPosted: Fri Aug 04, 2006 6:14 am Post subject: Reply with quote
OK, if I now 'predict' how teams do in the '06 playoffs, these expectations can be added to point differentials:

Code:
+2.8 Det
+2.5 Chi
+0.6 Was
+0.5 Cle
-0.3 Mia
-0.3 Ind
-1.4 NJ
-1.9 Mil

+1.1 Sac
+0.7 Dal
+0.5 SA
+0.0 LAL
-0.2 Phe
-0.3 Den
-2.1 Mem
-2.3 LAC


Numbers differ from previous chart because I've:
- converted 'production units' to 'points', roughly.
- used actual playoff mpg for players, rather than RS mpg
- normalized to add to zero (not sure why this step was necessary)

It looks like Sac-SA should be tighter than expected, by 0.6 ppg.
Give Dal an extra 2.8 vs Mem!
Den should get another 2.0 vs LAC -- caveat that neither team had much PO experience, Clipps in particular.
Lakes and Phx as expected.

Back east, Ind should trouble NJ by 1.1 ppg.
Was/Cle are as the oddsmakers would say.
Det should roll over Mil by almost 5 extra ppg.
Chi gains almost 3 on Mia.

Back to reality: This year, almost as many players changed their pattern as followed them. I'll go out on a limb and say this year was anomalous, and next year we'll see more continuity.

The Pistons and Spurs saw their fortunes largely reversed. The Cavs (non-Lebron) wilted. Kenyon crashed in the first turn. Wade fueled the Heat.
_________________
`
36% of all statistics are wrong
Back to top
View user's profile Send private message Send e-mail
Mike G



Joined: 14 Jan 2005
Posts: 3630
Location: Hendersonville, NC

PostPosted: Tue Oct 17, 2006 7:40 am Post subject: Reply with quote
Now that rosters have mostly stabilized, we can insert career PO/RS numbers next to this year's names. This playoff/season ratio is year-by-year, summed, and weighted by minutes in each year's playoffs.

Using a cutoff of at least 200 career playoff minutes, there are some 248 players in the league with this minimum. So I'll break the list into halves, going first with the Eastern Conference.

EDIT: I've divided player po/rs by the standard po/rs (.935) to get equal numbers of players above and below 1.00 . A ratio >1 means a playoff advantage to the team.

Code:
Team eG PO/RS Player

atl 14 1.00 johnson,joe
atl 12 1.03 Claxton
atl 10 .96 wright,lorenzen
atl 7 1.01 lue,tyronn

eG = 'equivalent games', or multiples of 36 minutes. This is a convenient shorthand for 'minutes/36'.

JJ is about at the league average of .935 po/rs; the median is quite a bit lower -- about .91 for players >200 min.

bos 42 1.11 pierce,paul
bos 27 .88 szczerbiak,wally
bos 14 .97 ratliff,theo
bos 6 .48 olowokandi,micha

Kandi is so bad, he can take your team out of the playoffs.

cha 7 .90 harrington,othel
cha 6 .83 knight,brevin

'Cats have more than doubled their playoff experience (if OH plays)

chi 93 1.02 wallace,ben
chi 62 .98 brown,p.j.
chi 18 1.18 griffin,adrian
chi 12 1.24 hinrich,kirk
chi 12 1.24 nocioni,andres
chi 12 .97 gordon,ben
chi 8 .81 duhon,chris
chi 8 1.14 allen,malik

The up-and-coming Bulls are stocked with up-and-comers.

cle 62 1.07 snow,eric
cle 54 .94 wesley,david
cle 25 1.05 jones,damon
cle 25 .95 hughes,larry
cle 19 .99 marshall,donyell
cle 18 .88 pollard,scot
cle 17 .91 james,lebron
cle 14 1.08 gooden,drew
cle 13 .87 ilgauskas,zydrun
cle 7 1.16 varejao,anderson

Lebron's 'career' po/rs is >50% vs the Pistons.

det 113 .99 wallace,rasheed
det 87 1.06 hamilton,richard
det 87 1.15 prince,tayshaun
det 79 1.03 billups,chauncey
det 55 .90 hunter,lindsey
det 32 1.12 mcdyess,antonio
det 20 1.07 mohammed,nazr
det 14 .83 murray,ronald
det 8 1.01 Cato

The only 'overachiever' last year was Hunter. Prince's ratio is inflated by his rookie playoff blossoming.

ind 50 .89 o'neal,jermaine
ind 41 .97 jackson,stephen
ind 33 .87 armstrong,darrel
ind 26 .94 tinsley,jamaal
ind 21 1.09 foster,jeff
ind 16 .96 daniels,marquis
ind 14 .84 harrington,al

Pacers are stocking up on known playoff dogs; even re-acquiring Harrington.

mia 210 1.02 o'neal,shaquille
mia 160 .95 payton,gary
mia 71 .97 mourning,alonzo
mia 70 .99 walker,antoine
mia 58 1.12 wade,dwyane
mia 42 .95 Williams, Jason
mia 40 .96 haslem,udonis
mia 28 1.18 posey,james
mia 8 .84 doleac,michael

Shaq is slipping, and Posey's tended to follow really weak seasons with less-weak playoffs. That leaves Dwyane to lead.

mil 14 .90 redd,michael
mil 11 .91 patterson,ruben

njn 113 .91 robinson,cliffor
njn 96 .98 kidd,jason
njn 63 .96 jefferson,richar
njn 41 .88 collins,jason
njn 34 1.09 Carter
njn 15 1.12 krstic,nenad
njn 6 .80 House

nyk 53 1.01 rose,jalen
nyk 36 .91 Rose, Malik
nyk 22 .87 marbury,stephon
nyk 18 1.22 James
nyk 16 .83 richardson,quent
nyk 12 .96 jeffries,jared
nyk 7 1.22 francis,steve

orl 31 .96 Turkoglu
orl 18 1.09 Battie
orl 14 .98 hill,grant
orl 13 .91 garrity,pat
orl 11 .75 outlaw,bo
orl 8 1.34 dooling,keyon
orl 6 .90 Arroyo

phi 76 1.05 iverson,allen
phi 70 .97 webber,chris
phi 14 .95 henderson,alan
phi 8 1.03 Ollie
phi 7 .80 hunter,steven

Webber almost cancels Ivy.

tor 72 1.09 Davis, Antonio
tor 27 .79 nesterovic,rasho
tor 19 .90 jones,fred
tor 7 .96 peterson,morris

was 38 1.15 daniels,antonio
was 25 1.33 butler,caron
was 21 .95 jamison,antawn
was 19 .98 arenas,gilbert
was 11 1.09 haywood,brendan



In general, the significance of the PO/RS column is in the size of the eG column. However, older players will tend to underachieve relative to their past. Players on the upside of their careers will tend to get better.

Further, if one can surmise how many minutes a guy will be getting in future playoffs, his PO/RS can be scaled to that.
_________________
`
36% of all statistics are wrong

Last edited by Mike G on Wed Oct 18, 2006 8:50 am; edited 2 times in total
Back to top
View user's profile Send private message Send e-mail
Mark



Joined: 20 Aug 2005
Posts: 807


PostPosted: Tue Oct 17, 2006 12:36 pm Post subject: Reply with quote
Career ratios probably should influence size of contract and trade value. I will look over the new lists. Heightened performances from Caron Butler, Posey and A Daniels catch my eye in addition to some of the better known ones. Thanks for sharing them.


(Looking back briefly at the data in the orginal post:
Maybe Piatowski interested Phoenix because of his playoff increase.
This is another reason I'd trade out Damon Jones though I guess he isnt really a major piece.And another argument for Anthony Johnson being a good pickup. Kobe the biggest faller. Teams able to curtail him a cinsiderable amount in the playoffs. Nash second biggest point faller. Vince Carter the biggest gainer. Might be fierce competiton to get him by one means or another. Tim Thomas 2nd biggest gainer. Single year data can be significantly influenced by quality of counterpart matchup.)
Back to top
View user's profile Send private message
Mike G



Joined: 14 Jan 2005
Posts: 3630
Location: Hendersonville, NC

PostPosted: Wed Oct 18, 2006 9:01 am Post subject: Reply with quote
Here's the West, same deal:
Code:

Team eG po/rs player
dal 82 .98 Nowitzki
dal 50 .94 Stackhouse
dal 41 .82 VanHorn
dal 37 1.01 Terry
dal 36 .95 george,devean
dal 35 1.03 howard,josh
dal 33 1.06 johnson,anthony
dal 32 .93 croshere,austin
dal 21 .98 dampier,erick
dal 17 .83 Harris
dal 15 .99 buckner,greg
dal 10 1.15 diop,desagana

A full roster of players with playoff history, and only a couple of regulars seem to be 'playoff guys'.

den 57 1.06 martin,kenyon
den 47 1.02 camby,marcus
den 36 .92 smith,joe
den 20 .89 najera,eduardo
den 16 1.08 miller,andre
den 15 .85 anthony,carmelo
den 11 1.05 boykins,earl
den 9 1.27 Carter
den 7 .99 evans,reggie

Doesn't look like a group that should fold up in the playoffs.

gsw 36 1.23 davis,baron

Baron's history is quite dated, of course.

hou 78 1.02 mutombo,dikembe
hou 28 1.05 mcgrady,tracy
hou 27 1.08 Wells
hou 14 .86 howard,juwan
hou 11 .97 Ming
hou 9 .68 battier,shane
hou 8 .80 alston,rafer

Mutombo has almost as many playoff minutes as everyone else.

lac 99 .97 Cassell
lac 43 1.19 Thomas
lac 28 .87 Mobley
lac 26 .97 Williams
lac 13 1.03 brand,elton
lac 10 1.26 Livingston
lac 9 1.47 Ross
lac 9 .99 Murray
lac 8 1.25 Maggette
lac 8 .95 Kaman

lal 134 .98 Bryant
lal 50 1.05 mckie,aaron
lal 22 .97 odom,lamar
lal 14 .97 radmanovic,vladi
lal 10 1.17 walton,luke
lal 9 1.21 brown,kwame
lal 8 .70 parker,smush

mem 78 1.00 jones,eddie
mem 45 1.02 stoudamire,damon
mem 24 .99 atkins,chucky
mem 14 .75 miller,mike

min 52 .99 garnett,kevin
min 14 1.25 hassell,trenton
min 12 .93 madsen,mark
min 10 .97 blount,mark
min 10 .82 davis,ricky
min 10 .79 james,mike
min 10 1.10 hudson,troy

nor 58 .95 Stojakovic
nor 30 .96 jackson,bobby
nor 16 .92 mason,desmond
nor 7 1.11 pargo,jannero
nor 7 .94 chandler,tyson
nor 7 .80 butler,rasual

pho 75 .96 nash,steve
pho 62 .93 marion,shawn
pho 32 1.00 bell,raja
pho 31 .94 thomas,kurt
pho 20 1.03 stoudemire,amare
pho 19 .89 barbosa,leandro
pho 17 1.07 diaw,boris
pho 16 .97 jones,james
pho 15 .97 Jones

por 26 .88 LaFrentz
por 26 1.02 Lenard
por 23 1.08 magloire,jamaal
por 6 1.23 randolph,zach
por 6 1.07 dixon,juan

sac 57 1.04 bibby,mike
sac 31 1.07 artest,ron
sac 29 .87 miller,brad
sac 29 .91 williamson,corli
sac 18 .95 thomas,kenny
sac 6 1.44 hart,jason

san 174 1.14 horry,robert
san 133 1.09 duncan,tim
san 80 .94 parker,tony
san 79 .91 bowen,bruce
san 76 .93 finley,michael
san 67 .98 Van Exel
san 62 1.13 ginobili,manu
san 37 .85 barry,brent
san 14 .91 vaughn,jacque
san 8 .67 Udrih

sea 39 1.11 allen,ray
sea 15 .89 lewis,rashard
sea 10 .84 ridnour,luke
sea 6 1.25 collison,nick

uta 85 1.12 fisher,derek
uta 49 .99 Ostertag
uta 19 .84 okur,mehmet
uta 10 .99 harpring,matt
uta 8 .85 kirilenko,andrei

_________________
`
36% of all statistics are wrong
Back to top
View user's profile Send private message Send e-mail
Mike G



Joined: 14 Jan 2005
Posts: 3630
Location: Hendersonville, NC

PostPosted: Tue Apr 17, 2007 11:15 am Post subject: Reply with quote
Now that at lEast one Conference has its playoff layout fixed, I'll make a first pass at amending playoff odds by figuring in past player PO/RS (playoff-to-regular-season production ratio) histories.

Regressing to the mean is important. This mean seems to be about a .945 PO/RS. The bigger a player's playoff history, the less he should regress. After age 31, players seem to drop off. Rather than delve into the formulas I've used, here are my guesses at teams' over/under -- relative to whatever Vegas or insider info one may possess:
Code:
3.1 Sag Dif net

Det 92.9 1.1 +2.2
Orl 89.6 -1.1

Chi 93.6 1.0 +4.7
NJ 89.9 -3.7

Cle 93.2 -1.7 -2.0
Mia 89.2 0.3

Tor 91.1 -1.5 -2.6
Was 88.5 1.1

EDIT: confusing column deleted; 'net' refers to advantage relative to favored (1st-listed) team, i.e. Tor, Cle, Chi, Det

Mmm, 3.1 ppg is the Sagarin home-court advantage. Sagarin's rating is listed, then the team's player-sum PO/RS advantage (or deficit, if negative).

Arenas and Butler are out, so the Wiz should be adjusted downward. I don't know what to sub for their Sag rating. . (Changed with Edit, see following post)

Here are players, by series, who are expected (by my formula) to over- or under-achieve by at least 0.5 PPG:
Code:
tm exp +/- player mpg
Det .60 2.1 Prince,Tayshaun 37
Det .64 1.0 Hamilton,Richard 38
Det .63 .6 Billups,Chauncey 37
Det .70 -.5 Wallace,Rasheed 34
Det .26 -.5 Murray,Ronald 20
Det .08 -.5 Delfino,Carlos 16
Det .56 -.8 Webber,Chris 30
Orl .20 .7 Dooling,Keyon 23
Orl .08 -.5 Milicic,Darko 23
Orl .28 -.5 Hill,Grant 31

Chi .26 1.7 Hinrich,Kirk 36
Chi .26 1.3 Nocioni,Andres 28
Chi .21 -.6 Duhon,Chris 25
Chi .17 -1.0 Deng,Luol 37
NJ .42 1.1 Carter,Vince 38
NJ .53 -.5 Jefferson,Richar 37
NJ .68 -.7 Robinson,Clifford 21
NJ .67 -.9 Kidd,Jason 37

Cle .28 .5 Gooden,Drew 28
Cle .27 -.9 Ilgauskas,Zydrun 28
Cle .30 -1.2 James,Lebron 41
Mia .51 2.6 Wade,Dwyane 39
Mia .34 .7 Posey,James 25
Mia .44 -.5 Williams,Jason 29
Mia .57 -.7 Mourning,Alonzo 23
Mia .80 -.7 Payton,Gary 24

Tor .36 -1.0 Nesterovic,Rasho 21
Was .08 .7 Stevenson,Deshawn 28
Was .44 .6 Daniels,Antonio 22
Was .33 -.5 Jamison,Antawn 38

'exp' is an Experience factor; basically, the % of a player's expectation which is based on his experience. Deshawn's 8% suggests he's 92% likely to achieve closer to the mean (.945) than to his mild history (37 min. at 1.48 PO/RS)

Krstic was one of the Nets' few PO/RS overachievers. I've figured he isn't playing, to get these adjustments. Anyone else out, from this player list?
_________________
`
36% of all statistics are wrong

Last edited by Mike G on Wed Apr 18, 2007 8:08 am; edited 3 times in total
Back to top
View user's profile Send private message Send e-mail
Mark



Joined: 20 Aug 2005
Posts: 807


PostPosted: Tue Apr 17, 2007 2:22 pm Post subject: Reply with quote
Is the last column net differential in effect for one game, neutral assumption on home court?

Could you take home differentials, who has home court advantage and series schedule and compute expected wins of each game and then expected win % of series?
Back to top
View user's profile Send private message
Mike G



Joined: 14 Jan 2005
Posts: 3630
Location: Hendersonville, NC

PostPosted: Wed Apr 18, 2007 7:57 am Post subject: Reply with quote
Mark wrote:
Is the last column net differential in effect for one game, neutral assumption on home court?

Could you take home differentials, who has home court advantage and series schedule and compute expected wins of each game and then expected win % of series?


Sorry about the confusion. Yes, that's what to do. As I mentioned, it's up to the reader to recalibrate based on missing players, teams due to peak (Mia?), etc.

Ay-yi, I forgot Caron Butler was also out. Without him and Gil, the Wiz have only +1.1 differential from expected. More importantly, they are the worst team in the NBA, by far. They've played 8 games, all against East teams, with a PtDiff of -6.5. This projects to a 20-win season -- if all 82 games were in the East and 5/8 were at home.
_________________
`
36% of all statistics are wrong
Back to top
View user's profile Send private message Send e-mail
Mike G



Joined: 14 Jan 2005
Posts: 3630
Location: Hendersonville, NC

PostPosted: Sat Apr 21, 2007 6:21 am Post subject: Reply with quote
Western Conference players expected to over- or underachieve in playoffs by at least 0.5 PPG either way. Based on playing time, playoff experience, age (years over 31), and of course career PO/RS productivity ratios.

Columns show over/under, season mpg, an experience factor (smaller number regresses more to the mean). Distributed by 1st-round series:
Code:
over mpg exp player hist team
.7 35 .42 Howard,Josh .97 Dal
-.5 23 .47 Stackhouse,Jerry .88 Dal
-.9 26 .30 Harris,Devin .77 Dal
3.2 38 .42 Davis,Baron 1.15 GS
-.8 34 .29 Harrington,Al .79 GS

3.5 35 .75 Duncan,Tim 1.02 SA
2.3 28 .52 Ginobili,Manu 1.05 SA
1.1 17 .85 Horry,Robert 1.07 SA
-.5 14 .21 Udrih,Beno .62 SA
-.8 31 .60 Bowen,Bruce .85 SA
-.9 34 .60 Parker,Tony .88 SA
-1.1 22 .41 Barry,Brent .79 SA
1.4 43 .59 Iverson,Allen .98 Den
-1.4 39 .28 Anthony,Carmelo .79 Den

1.0 36 .38 Mcgrady,Tracy .98 Hou
-.8 27 .28 Howard,Juwan .80 Hou
-1.0 37 .21 Alston,Rafer .75 Hou
-1.5 37 .22 Battier,Shane .64 Hou
1.1 27 .61 Fisher,Derek 1.04 Uta
-.5 17 .19 Giricek,Gordan .60 Uta
-.6 31 .21 Kirilenko,Andrei .79 Uta
-1.2 34 .29 Okur,Mehmet .78 Uta

.6 33 .34 Diaw,Boris 1.00 Pho
.5 32 .34 Stoudemire,Amare .97 Pho
-.7 32 .33 Barbosa,Leandro .83 Pho
-.9 35 .61 Nash,Steve .89 Pho
-1.0 38 .53 Marion,Shawn .87 Pho
1.0 33 .24 Walton,Luke 1.10 LAL
.9 28 .21 Brown,Kwame 1.13 LAL
.7 22 .16 Evans,Maurice 1.33 LAL
-1.0 31 .20 Parker,Smush .65 LAL

In 158 playoff minutes, Maurice Evans has produced at 133% of his RS rates. This smallish sample is regressed to 107% 'expected' PO/RS. For the West, the 'mean' expectation is only .92 PO/RS. That's just the figure that makes these sum to zero:

Code:
team Sagar over Net Expec
Dal 98.1 -.5 -1.8 5.9
GS 90.4 1.4

SA 98.4 2.6 3.4 9.7
Den 92.01 -.7

Hou 94.6 -1.6 -0.7 1.2
Uta 92.7 -.9

Pho 96.9 -1.4 -2.5 3.9
LAL 90.6 1.1

'Net' and 'Expected' refer to the favored team (Tm1). Net is = over1 - over2 ; Expec is = Sag1 - Sag2 + Net

Suns/Lakers should be 2.5 PPG tighter than expected (by RS records). With a 3.1 PPG home court advantage, games in LA are almost even.

Spurs are loaded with 'playoff guys' and should roll.
_________________
`
36% of all statistics are wrong
Back to top
View user's profile Send private message Send e-mail
Mike G



Joined: 14 Jan 2005
Posts: 3630
Location: Hendersonville, NC

PostPosted: Sun Apr 22, 2007 7:04 am Post subject: Reply with quote
Here are the actual first round matchups, for East and West; these 'differentials' can be carried to later rounds, since they are relative. Just subtract any 2 teams' adjusted rankings to get the favorite and the spread.
Code:
East Sagar Dif AdjSa Fave
Det 92.9 1.1 94.0 5.4
Orl 89.6 -1.1 88.6

Chi 93.6 1.0 94.6 5.1
Mia 89.2 0.3 89.5

Cle 93.2 -1.7 91.5 11.5
Was 80.0 0.0 80.0

Tor 91.1 -1.5 89.6 3.4
NJ 89.9 -3.7 86.2

West Sagar Dif AdjSa Fave
Dal 98.1 -.5 97.6 5.9
GS 90.4 1.4 91.8

SA 98.4 2.6 101.0 9.7
Den 92.1 -.7 91.3

Hou 94.6 -1.6 93.0 1.2
Uta 92.7 -.9 91.8

Phx 96.9 -1.4 95.6 3.9
LAL 90.6 1.1 91.7

_________________
`
36% of all statistics are wrong

horsecow



Joined: 01 Dec 2009
Posts: 12
Location: Iowa City, IA

PostPosted: Fri May 28, 2010 3:46 pm Post subject: Championship equivalents using WinShares Reply with quote
Hi all - first-time poster here. After some people tweeted how it's a shame Steve Nash will probably never win a ring, I decided to write a little entry on how many "rings" Steve Nash would have if we converted his playoff WinShares into championship-equivalent blocks of wins. I did the same for the top 101 players in WinShares. The only reason it wasn't straightforward was that the NBA has changed the length of the playoffs several times over the years, so I had to adjust for that. The results are as you'd expect, with some players from the shorter playoff era moving up and newer players moving down. And Steve Nash? He's accounted for .73 of a championship all by himself, which isn't too bad, and is miles ahead of Adam Morrison.

The full post is at http://jamerchant.wordpress.com/2010/05 ... /#more-382

Here's the top ten (the full list is at the website -- I won't clog things up here):

Bill Russell || 2.81 championship equivalents
Kareem Abdul-Jabbar || 2.73
Wilt Chamberlain || 2.70
Michael Jordan || 2.65
Jerry West || 2.43
Magic Johnson || 2.26
Shaquille O’Neal || 2.03
Tim Duncan || 1.82
Larry Bird || 1.72
John Havlicek || 1.65
Back to top
View user's profile Send private message
DSMok1



Joined: 05 Aug 2009
Posts: 611
Location: Where the wind comes sweeping down the plains

PostPosted: Fri May 28, 2010 4:08 pm Post subject: Re: Championship equivalents using WinShares Reply with quote
horsecow wrote:
Hi all - first-time poster here. After some people tweeted how it's a shame Steve Nash will probably never win a ring, I decided to write a little entry on how many "rings" Steve Nash would have if we converted his playoff WinShares into championship-equivalent blocks of wins. I did the same for the top 101 players in WinShares. The only reason it wasn't straightforward was that the NBA has changed the length of the playoffs several times over the years, so I had to adjust for that. The results are as you'd expect, with some players from the shorter playoff era moving up and newer players moving down. And Steve Nash? He's accounted for .73 of a championship all by himself, which isn't too bad, and is miles ahead of Adam Morrison.

The full post is at http://jamerchant.wordpress.com/2010/05 ... /#more-382

Here's the top ten (the full list is at the website -- I won't clog things up here):

Bill Russell || 2.81 championship equivalents
Kareem Abdul-Jabbar || 2.73
Wilt Chamberlain || 2.70
Michael Jordan || 2.65
Jerry West || 2.43
Magic Johnson || 2.26
Shaquille O’Neal || 2.03
Tim Duncan || 1.82
Larry Bird || 1.72
John Havlicek || 1.65


Clever, and a good use of win shares!
Back to top
View user's profile Send private message Send e-mail Visit poster's website
Mike G



Joined: 14 Jan 2005
Posts: 3597
Location: Hendersonville, NC

PostPosted: Fri May 28, 2010 5:29 pm Post subject: Reply with quote
Yes, this is a bold try at cross-era comparisons.
This little snippet jumped out at me:
Code:
player rings rank WS rank
Frank Ramsey* 1.05 35 9.49 75
Bill Sharman* 1.05 36 9.32 81
Bob Cousy* 1.04 37 9.38 80

Normally, one would expect a reverse order for these 3 teammates, and rather strongly. Cousy was the multiple all-world guy, Ramsey came off the bench most of the time.

But in playoffs, Ramsey was something of a Ginobili in his time.
Cousy probably gets socked by Win Shares for his low shooting%.

It was easier to win a title when there were 2 rounds to the playoffs -- most of Russell's titles were this sort. Is there any way to make a 4-round title worth more, but not twice as much? Maybe take the square root?

Consider that in each of Jordan/Pippen's 3-peats, they outlasted (3x15) 45 playoff rivals. In Russell's 9-straight, they prevailed against (9x5) 45, also?

If you do it by # of teams in the league, it favors the Bulls even more.
_________________
`
36% of all statistics are wrong
Back to top
View user's profile Send private message Send e-mail
horsecow



Joined: 01 Dec 2009
Posts: 12
Location: Iowa City, IA

PostPosted: Sat May 29, 2010 10:10 am Post subject: Reply with quote
That's a good point about the playoffs being a harder slog in the modern era. I guess the rejoinder would be that the league was less diluted in the old days, so you wouldn't have a first round match-up with the Charlotte Hornets to pad your stats.
Back to top
View user's profile Send private message
Mike G



Joined: 14 Jan 2005
Posts: 3597
Location: Hendersonville, NC

PostPosted: Sat May 29, 2010 11:20 am Post subject: Reply with quote
That is a fairly common misconception. When 6 of 8/9 teams get into the playoffs, there are more mediocre teams in the playoffs. Several years there were only 3 teams above .500 .

The Celtics' first title was gained by beating 38-34 Syracuse, then 34-38 St Louis. They won against totally average competition.

Next year, StL advanced to the Finals by beating 33-39 Cincinnati.

To win the first of their 8-straight (1959-66), the Celts had to overcome .486 Syr and the .458 Lakers.

The worst Charlotte Hornets team to make the playoffs was 44-38 (.537). This record would have led the Western Division a couple times in Pettit's day.

One or two additional rounds, even against the occasional sub-.500 team, cannot make it easier to win a title. It's another 2 chances to be upset. (Ask Dirk.)
_________________
`
36% of all statistics are wrong

Crow
Post subject: Pace vs. Efficiency
PostPosted: Wed Apr 20, 2011 1:20 pm
Offline

Joined: Thu Apr 14, 2011 7:10 pm
Posts: 403
Author Message
wmchad



Joined: 16 Apr 2009
Posts: 13
Location: Portland, OR

PostPosted: Thu Apr 30, 2009 3:14 pm Post subject: Pace vs. Efficiency Reply with quote
Hello. I'm relatively new to the board, so please forgive my lack of knowledge, etiquette...

Has anyone looked at the relationship between pace and efficiency for teams? I've seen the thread on usage vs efficiency for players and was wondering if something similar could be seen on a team level. I would imagine that, if teams had an optimal pace, you would see a drop-off in efficiency as they moved away from that pace.

I've tried to do some regressions myself for this regular season, but I don't know how to interpret the results. For example, the highest r^2 for home defensive efficiency vs pace is Dallas: pace coefficient is -0.89, r^2 = 0.235. I tried a number of variations, including using a pace^2 term and trying to control for opponent strength, and my highest r^2 was 0.300. Has anyone else done a similar study?

Thank you,
Chad Young
Back to top
View user's profile Send private message
kjb



Joined: 03 Jan 2005
Posts: 865
Location: Washington, DC

PostPosted: Thu Apr 30, 2009 3:46 pm Post subject: Reply with quote
I've done very limited looks at pace vs. efficiency. Generally, there seems to be miniscule effects on efficiency. If I'm recalling the numbers correctly, better defensive teams are a bit slower paced on average. But even that effect is pretty small, I think.
Back to top
View user's profile Send private message AIM Address Yahoo Messenger
bchaikin



Joined: 27 Jan 2005
Posts: 688
Location: cleveland, ohio

PostPosted: Thu Apr 30, 2009 8:44 pm Post subject: Reply with quote
Has anyone looked at the relationship between pace and efficiency for teams?

looking at just some extremes, in this case the annual off/def pts/poss of the 5 teams with the fastest game pace versus the 5 teams with the slowest game pace over the past 30 seasons of the nba (since 1979-80):

column A - avg off pts/poss scored of the 5 teams with the fastest game pace
column B - avg def pts/poss allowed of the 5 teams with the fastest game pace
column C - avg off pts/poss scored of the 5 teams with the slowest game pace
column D - avg def pts/poss allowed of the 5 teams with the slowest game pace

--year--------A---------B---------C--------D--
197980---1.037---1.056---1.036---1.052
198081---1.049---1.033---1.039---1.068
198182---1.069---1.070---1.047---1.062
198283---1.042---1.055---1.009---1.012
198384---1.083---1.083---1.050---1.057
198485---1.060---1.067---1.054---1.074
198586---1.062---1.068---1.036---1.053
198687---1.057---1.072---1.083---1.056
198788---1.069---1.074---1.080---1.059
198889---1.081---1.055---1.074---1.037
198990---1.051---1.076---1.061---1.055
199091---1.060---1.075---1.063---1.061
199192---1.081---1.067---1.055---1.063
199293---1.071---1.070---1.079---1.063
199394---1.049---1.063---1.065---1.018
199495---1.104---1.078---1.064---1.039
199596---1.059---1.068---1.087---1.052
199697---1.041---1.090---1.055---1.032
199798---1.039---1.061---1.044---1.010
199899---1.011---1.016---1.021---0.991
199900---1.020---1.033---1.022---1.009
200001---1.017---1.017---1.016---0.999
200102---1.058---1.037---1.011---1.025
200203---1.020---1.039---1.013---1.015
200304---1.048---1.037---1.004---0.999
200405---1.059---1.047---1.041---1.036
200506---1.046---1.044---1.043---1.029
200607---1.078---1.065---1.066---1.035
200708---1.073---1.072---1.082---1.047
200809---1.087---1.091---1.076---1.056

average--1.057---1.060---1.050---1.040

you find at the extremes a slight advantage for higher offensive efficiency for the fastest paced teams annually, and a more pronounced advantage defensively for the slowest paced teams annually...

notice that there are 12 seasons where the 5 slowest game paced teams had an average off pts/poss scored that was actually higher than the 5 fastest game paced teams, including the five year stretch of 1995-96 to 1999-00..

but in only 2 seasons (1980-81 and 1984-85) did the 5 fastest paced teams have an average def pts/poss allowed that was less than the avg def pts/poss allowed of the 5 slowest paced teams...
Back to top
View user's profile Send private message Send e-mail Visit poster's website
Mountain



Joined: 13 Mar 2007
Posts: 1527


PostPosted: Fri May 01, 2009 2:52 am Post subject: Reply with quote
For the fast 5 offensive efficiency in 2008-9 was second best for the period compared to the other year's fast 5s but defensive efficiency was the worst in the group. For the slowest 5 I think it is 6th best on offense and 9th worst on defense.

The slow 5 had its 3rd biggest net efficiency spread in 2007-8 but lost about 40% of it this season.

The fast 5 have been negative on net about 2/3rds of the time. Last season wasn't far different from median but was a bit better compared to how the fast 5 have faired before.



If you look at the below mean pace teams, there were 9 who were above average on 3 pt makes, 8 made playoffs and I think 4 will advance.

Cavs were 2nd slowest, 1st on 3 pt makes within this group and 1st on defensive efficiency within the group. They are the standard for this consciously constructed style.


Below average pace and less average on 3 pt makes there were 7 and 3 made playoffs and up to 2 may advance.

Celtics are the class of this group. 4th on pace, 2nd on 3 pt makes, 1st on defensive efficiency.


For fast pace teams with above average 3 pt makes there were 8 only 3 made playoffs but at least 2 will advance.

Lakers were 4th in group on pace, 2nd on defensive effiiciency but last on 3 pt makes in that group.


For fast pace teams with below average 3 pt makes there were 6, 2 made playoffs, 1 still could advance- Chicago.

The last group was least successful- this season.


Odds are the champ will come from the first or third group- each with above mean 3 pt makes and almost certainly strong defense. Pace is the perhaps the weak partner in this threesome.

Last season Boston was the 2nd slowest among those above average on 3 pt makes, 2nd best on 3 pt makes in the group and best on defensive efficiency. Real close to what the Cavs are now. That is the formula I'd prefer / use if you have the coach & talent for it.

Offensive efficiency is obviously important but I wanted to put the focus on the pace- 3 pt - defensive efficiency relationship.

Last edited by Mountain on Fri May 01, 2009 11:38 pm; edited 2 times in total
Back to top
View user's profile Send private message
gabefarkas



Joined: 31 Dec 2004
Posts: 1313
Location: Durham, NC

PostPosted: Fri May 01, 2009 7:33 am Post subject: Reply with quote
Are you asking about teams compared to other teams, or within-team fluctuations game-to-game?

For example, are you trying to (hypothetically) say that Team A, who plays at a fast pace, has a better Off Eff than Team B, who plays at a slower pace? Or are you trying to say that Team A has a better Off Eff when playing in games at a faster pace than it does when playing in games at a slower pace?
Back to top
View user's profile Send private message Send e-mail AIM Address
wmchad



Joined: 16 Apr 2009
Posts: 13
Location: Portland, OR

PostPosted: Fri May 01, 2009 10:22 am Post subject: Reply with quote
Quote:

Are you asking about teams compared to other teams, or within-team fluctuations game-to-game?


I was specifically asking about within-team fluctuations. I know there are a lot of factors, but I thought that if a team has a 'natural' pace, then they would prefer to play at that pace and you would see fall off in efficiency as they deviated from that pace. I tried looking at offensive and defensive efficiency as well as efficiency differential. I also tried adjusting for opponent strength.

As an example, here is the Mavericks' home defensive efficiency for the season.



As opposed to the same graph for the Clippers.


Back to top
View user's profile Send private message
Mountain



Joined: 13 Mar 2007
Posts: 1527


PostPosted: Fri May 01, 2009 11:40 pm Post subject: Reply with quote
Charts like this are useful. More useful than just the averages.




I see only 1.5 teams on average play above 94 average pace in the last 6 playoffs. Less than 10%. This regular season 23% played at least this fast; in playoffs none so far. You better be able to win slower than real fast.
Back to top
View user's profile Send private message
Mike G



Joined: 14 Jan 2005
Posts: 3597
Location: Hendersonville, NC

PostPosted: Sat May 02, 2009 5:48 am Post subject: Reply with quote
Any reason a team would have more of a preferred pace at home (vs away)? Home+Away gives you twice the sample size.
_________________
`
36% of all statistics are wrong
Back to top
View user's profile Send private message Send e-mail
wmchad



Joined: 16 Apr 2009
Posts: 13
Location: Portland, OR

PostPosted: Sat May 02, 2009 7:35 am Post subject: Reply with quote
I mainly separated them out because I wasn't adjusting for the home/away difference in efficiency. I may try running them again for all games with the adjustment to get the larger sample.
Back to top
View user's profile Send private message
kjb



Joined: 03 Jan 2005
Posts: 865
Location: Washington, DC

PostPosted: Mon May 04, 2009 12:59 pm Post subject: Reply with quote
What r^2 are you getting on those charts? When I generate similar charts in Excel for the Wizards, I'm getting 0.0032 on ortg and 0.0004 on the drtg. That's when I do the linear trendline. I get highest r^2 when I use the polynomial trendline with order set to 6. Those are still very low.

In fact, I get the strongest r^2 for this sort of thing when I first calculate the team's average performance by pace. So, if they have 6 games at a pace factor of 90, then I just average those together. Then I create the same chart using the polynomial trendline with order set to 6.

Here's what it looks like:

Back to top
View user's profile Send private message AIM Address Yahoo Messenger
wmchad



Joined: 16 Apr 2009
Posts: 13
Location: Portland, OR

PostPosted: Wed May 06, 2009 12:05 pm Post subject: Reply with quote
The r^2 for the Dallas chart was 0.235. For the Clippers chart it was basically 0 (< 0.0001). I'm on vacation so I can't look up what I had for the Wizards, but they could have been anything in between those.

I used pace and pace^2 (polynomial order 2) regressions because I couldn't think of a reason that the data might respond to a higher order than that. A polynomial regression will always have an r^2 equal to or higher than another with a lower order polynomial, so that's why you get the highest r^2 when the order is 6.

Also, averaging games at a certain pace will increase your r^2 because you're removing outliers. I have considered looking at a running local average (finding the average efficiency at a pace +/- 2 possessions). Just to see what the data looks like.

When I get back from vacation I can look up what I have for the Wizards if you want. Looking at your data, you might want to subtract offensive efficiency from defensive efficiency (by pace) and see what you come up with.
Back to top
View user's profile Send private message
wmchad



Joined: 16 Apr 2009
Posts: 13
Location: Portland, OR

PostPosted: Wed May 13, 2009 6:41 pm Post subject: Reply with quote
I looked up the data I have for the Wizards, and I get low r^2s for them across the board. The highest I have is 0.020 for home defensive efficiency, and none of the regressions have a t value > 1 (from what I understand, higher t value means greater confidence that something is significant - I think you want it to be > 2). Contrast this with the Dallas results I put up earlier, where the r^2 is 0.235 and the t value is ~ 3.5. As far as I know (and I'm just starting to learn here, so please correct me if I'm wrong), what I'm seeing indicates that pace does not have an observable effect on the Wizard's efficiency, whereas it appears to have at least some effect for the Mavericks - at least in the case of home defensive efficiency.

Charles



Joined: 16 May 2005
Posts: 134


PostPosted: Fri Jun 13, 2008 2:26 pm Post subject: Are large market playoff teams favored? Reply with quote
The NBA is a very high profile business, so protecting the integrity of its product is paramount. It is clearly in the league's own best interest to be seen as 100% impartial. However, the league also has an undeniable interest in maximizing media coverage and viewership. So, given the recent statements from Tim Donaghy, some subtle bias in favor of large market teams is a legitimate concern.

As a quick and dirty test, I looked at each playoff game over the past twelve seasons and took the difference between a team's regular season point differential and their opponent's point differential. I adjusted for home court advantage. Then I compared the game result to the "expected" result.

For example, suppose San Antonio is visiting Utah and the Spurs have a regular season point differential of +7.2 while the Jazz have a regular season differential of +1.2. The Spurs are expected to win by six points. However, since the Spurs are on the road we subtract 4.2 from their advantage for a final spread of +1.8. If San Antonio wins the game by four points then they performed +2.2 better than expected. We do that calculation for each of the Spurs 133 playoff games and get an average difference of +0.4 (within the expected range.)

Code:
Two teams have done much better in the playoffs than their regular season records suggested. The teams from the league's two largest cities -- New York and Los Angeles.

Unexplained Playoff Advantage
+2.4 NYK
+2.2 LAL

Then comes a group of four team with a one to two point "unexplained" edge -- mostly from mid-large markets. Btw, Toronto is North America's fifth largest city after Mexico City, New York, Los Angeles, and Chicago.

Unexplained Playoff Advantage
+1.7 CHI
+1.3 TOR
+1.2 BOS
+1.2 CHH

And, the rest of the teams.

Unexplained Playoff Advantage
+0.6 DET
+0.6 HOU
+0.4 SAS
+0.1 POR
+0.0 UTA
-0.1 DAL
-0.1 NJN
-0.3 IND
-0.5 MIL
-0.6 SAC
-0.9 PHI
-1.0 MIA
-1.2 ATL
-1.4 SEA
-1.8 PHO
-2.0 MIN
-2.2 ORL
-3.1 CLE

Are large market teams being favored or is there some other explanation?
Back to top
View user's profile Send private message
Brian M



Joined: 25 Nov 2006
Posts: 40


PostPosted: Fri Jun 13, 2008 3:47 pm Post subject: Reply with quote
I don't know how much it affects the data, but in 1999 the Knicks were objectively much better in the playoffs than in the regular season. They underwent a significant overhaul by bringing in Camby and Sprewell, there were only 50 or so games for the team to gel due to the lockout, and Van Gundy didn't really begin trusting the new guys until late in the season. One might legitimately question Larry Johnson's 4 point play, but on the whole I'd expect the Knicks to look much better statistically in those playoffs than they did in the regular season, and not because of a refereeing bias.
Back to top
View user's profile Send private message
Mountain



Joined: 13 Mar 2007
Posts: 1527


PostPosted: Fri Jun 13, 2008 5:14 pm Post subject: Reply with quote
This is a timely start of a study.

I'd be interested in

1) how much of the big market edge was coming in home vs road games,

2) if there is a strong correlation of "unexplained playoff advantage" with free throw differential or FT/FGA game by game for the large market teams,

3) if there is a strong correlation of "unexplained playoff advantage" with opponent market size, and

4) what the regular season differential was between each playoff match-up and how much "unexplained playoff advantage" there is compared to that.


Referee bias is just one possible explanation. I wonder how much big market total team salary, or ability to attract free agents major and minor and retain players, or good fortune with or ability to pay / retain the best coaches and GMs may contribute to it as well.

A 12 year study is a decent length but does a 20+ year study show something similar. less or more pronounced?
If you split between pre and post modern salary cap system
what is the direction of the advantage?
Back to top
View user's profile Send private message
Ryan J. Parker



Joined: 23 Mar 2007
Posts: 711
Location: Raleigh, NC

PostPosted: Sat Jun 14, 2008 8:04 am Post subject: Reply with quote
Help us understand what we're talking about here... What are the odds a +2.4 or +2.2 differential happens by random chance? Just how unlikely is this, assuming everything else being equal?
Back to top
View user's profile Send private message Visit poster's website
Charles



Joined: 16 May 2005
Posts: 134


PostPosted: Sat Jun 14, 2008 9:56 pm Post subject: Reply with quote
Brian M wrote:
I don't know how much it affects the data, but in 1999 the Knicks were objectively much better in the playoffs than in the regular season. They underwent a significant overhaul by bringing in Camby and Sprewell, there were only 50 or so games for the team to gel due to the lockout, and Van Gundy didn't really begin trusting the new guys until late in the season. One might legitimately question Larry Johnson's 4 point play, but on the whole I'd expect the Knicks to look much better statistically in those playoffs than they did in the regular season, and not because of a refereeing bias.

Sounds reasonable. The samples are fairly small, so I would certainly interpret the results cautiously when it comes to individual teams.
Back to top
View user's profile Send private message
Charles



Joined: 16 May 2005
Posts: 134


PostPosted: Sat Jun 14, 2008 10:03 pm Post subject: Reply with quote
Mountain wrote:

1) how much of the big market edge was coming in home vs road games,


The samples are not large to start with, so breaking individual team records down further is probably not a good idea. However, the three top big market teams -- Knicks, Lakers and Bulls all showed considerable improvement from the regular season in both home and road playoff games. As a group...

Unexplained Playoff Advantage: NYK, LAL, CHI
Home +2.28 (141 games)
Road +1.92 (135 games)


Mountain wrote:
Referee bias is just one possible explanation. I wonder how much big market total team salary, or ability to attract free agents major and minor and retain players, or good fortune with or ability to pay / retain the best coaches and GMs may contribute to it as well.

A 12 year study is a decent length but does a 20+ year study show something similar. less or more pronounced?
If you split between pre and post modern salary cap system
what is the direction of the advantage?


I am not suggesting it was referee bias. I am simply noting that large market teams have done better in the playoffs than their regular season records would have predicted. But, frankly, I don't understand why would you expect salary or any of the factors you mention to cause a team's scoring differential to increase from the regular season to the playoffs. Teams like Dallas and Phoenix have spent freely, retained excellent coaches but not seen their playoff results improve.
Back to top
View user's profile Send private message
Charles



Joined: 16 May 2005
Posts: 134


PostPosted: Sat Jun 14, 2008 10:06 pm Post subject: Reply with quote
Ryan J. Parker wrote:
Help us understand what we're talking about here... What are the odds a +2.4 or +2.2 differential happens by random chance? Just how unlikely is this, assuming everything else being equal?


Interesting. An average increase of two points over expectations for 100+ games seems pretty substantial, but I didn't actually look at that.

Perhaps the more pertinent question is: what is the chance of the four teams from the largest markets being the exact same teams whose point differential improved the most during the playoffs? Even ignoring the fact that they fall in the correct order the chances of that happening randomly would be 4/30 * 3/29 * 2/28 * 1/27, So, about one in 25,000.
Back to top
View user's profile Send private message
Mountain



Joined: 13 Mar 2007
Posts: 1527


PostPosted: Sat Jun 14, 2008 11:21 pm Post subject: Reply with quote
The Bulls large market contributed to Jordan's superstar call and no-call patterns as it did for others. It helped support his very expensive salary in the last part of his run there- #1 on team salary in the league the final 2 years.

New York being big stage New York definitely played a role in attracting Riley (at top dollar) and brought with him Jeff VanGundy. (putting aside the Ewing draft issue). And for the 2 years that ended in appearances in Finals they were #1 and #2 on total team salary. Supported by a big market.

Big market LA and the big check there helped bring Phil Jackson and Shaq (and Charlotte's market size influenced Kobe's draft day posture). Total team salary was 4th, 6th and 12th in the 3peat years. And recently made luxury tax Gasol more feasible than for other teams.


And if the big revenue streams bought a more expensive set of quality players and coaches I am not that surprised that the highest expression of that quality comes out in the playoffs... after the playing of the 82 game prelude. Of course not every big market team got max advantage via spending related influences but it is part of the picture.

The fact that the very biggest markets show the biggest playoff performance gains is notable and does raise the question if market size / "importance" played some role as well.

By referee impact .... or maybe some boost from fan support, team swagger & intimidation? Plenty of factors contributing or arguably contributing.

Last edited by Mountain on Mon Aug 04, 2008 8:22 pm; edited 1 time in total
Back to top
View user's profile Send private message
sptsjunkie



Joined: 17 Feb 2008
Posts: 16
Location: Los Angeles

PostPosted: Mon Aug 04, 2008 7:00 pm Post subject: Reply with quote
As a Kings fan, I would love to believe it was all of the evil refs. However, another factor that probably inflates at least LA and Houston having higher than expected post-season results is that both of those teams were notorious for starting off the season slowly and saving energy and then getting in a rhythm and peaking right as the playoffs approached.

Is there a simialr correlation if you look at how teams performed in the final 1/3 of the regular season and the playoffs?

Ed Küpfer



Joined: 30 Dec 2004
Posts: 784
Location: Toronto

PostPosted: Wed Sep 28, 2005 4:37 pm Post subject: Coaches playing rookies Reply with quote
Generalising from an earlier post attempting to estimate the effect of individual coaches on rookie minutes. I regressed every BAA, ABA, and NBA rookie's minutes played for each team-season against an assortment of relevant factors, including the coaching presence of the 25 coaches with the most games coached. (I really need to type "coach" more.)

Code:
Predictor Coef (SE Coef) Explanation
Constant -89 (56)
Top5 335** (59) Dummy variable, 1 if player was a top 5 draft pick, 0 if not.
Top10 417** (52) " " 1 if player was a top 10 draft pick, 0 if not.
1stRnd 454** (34) " " 1 if player was a first round draft pick, 0 if not.
ABA 452** (50) " " 1 if player-season was in the ABA, 0 if not.
ABAEra 129** (38) " " 1 if player-season was in the ABA Era (1968-76), 0 if not.

PER 37** (4) Player Efficiency Rating for that team-season.

Red Auerbach -349** (86) Dummy variable indicating whether coach in first column coached at least 41 games of the team-season
George Karl -144 (119)
Phil Jackson -129 (129)
Larry Brown -111 (85)
Chuck Daly -72 (131)
Pat Riley -45 (102)
Doug Moe -42 (129)
Rick Adelman -31 (128)
Red Holzman -24 (97)
Gene Shue 5 (83)
Lenny Wilkens 11 (77)
Dick Motta 47 (85)
Don Nelson 52 (82)
Slick Leonard 76 (106)
Jerry Sloan 79 (97)
Bill Fitch 86 (87)
Del Harris 122 (135)
Jack Ramsay 126 (93)
Al Attles 171 (116)
Cotton Fitzsimmons 172 (93)
Alex Hannum 190 (110)
Mike Fratello 198 (117)
Hubie Brown 243* (105)
Kevin Loughery 321** (120)
John MacLeod 349** (100)


What this means for you

The coaches above are listed in increasing effect on rookie minutes played. On average, Red Auerbach's rookies played 349 fewer minutes over the course of the season than "expected" given the other factors listed above. Only the numbers with asterisks nest to them are considered to be statistically significant. Larry Brown is on the edge of significance, depending on the variables you choose to regress.
_________________
ed
Back to top
View user's profile Send private message Send e-mail
Ed Küpfer



Joined: 30 Dec 2004
Posts: 784
Location: Toronto

PostPosted: Wed Sep 28, 2005 5:16 pm Post subject: Reply with quote
It occured to me that I may be doing this wrong—particularly the inclusion of some of the Draft Pick variables. PER I used as a proxy for general player ability, on the assumption that a player should get more minutes if he's playing well. I used PER rather than any other performance metrics because B-Ref's PER hack goes back further in time than the others. The ABA variables need to be there, I think, becuase the enviroment for players' minutes was so different.

Anything else I missed? Any other coaches who have a rap for dissing the rooks?
_________________
ed
Back to top
View user's profile Send private message Send e-mail
mateo82



Joined: 06 Aug 2005
Posts: 211


PostPosted: Wed Sep 28, 2005 6:02 pm Post subject: Reply with quote
I don't understand why you are making this so complicated. All you have to do is average rookie MPG over a coach's career. Why is PER factored in?
Back to top
View user's profile Send private message
Ed Küpfer



Joined: 30 Dec 2004
Posts: 784
Location: Toronto

PostPosted: Wed Sep 28, 2005 6:20 pm Post subject: Reply with quote
mateo82 wrote:
I don't understand why you are making this so complicated. All you have to do is average rookie MPG over a coach's career. Why is PER factored in?


Because a god player deserves more playing time than a bad player. You wouldn't fault Larry Brown for not playing his rookies if his rookies were Duane Ferrell or Antonio Lang. But you would fault him if his rookies were Hakeem or Duncan. The quality of the rookie needs to be taken into account before you decide whether they are being underutilised.
_________________
ed
Back to top
View user's profile Send private message Send e-mail
mateo82



Joined: 06 Aug 2005
Posts: 211


PostPosted: Wed Sep 28, 2005 6:38 pm Post subject: Reply with quote
Ed Küpfer wrote:


Because a god player deserves more playing time than a bad player. You wouldn't fault Larry Brown for not playing his rookies if his rookies were Duane Ferrell or Antonio Lang. But you would fault him if his rookies were Hakeem or Duncan. The quality of the rookie needs to be taken into account before you decide whether they are being underutilised.


How can you tell if a player is good unless they get quality minutes? You're assuming that a rookie who gets 15 mpg should be able to play up to his capabilities, as should a player who plays 30mpg or 45mpg.

Wouldn't it be easier to take a list of all of the rookies a coach has had and then manually pluck out the bad players (players who were bad on multiple stops)? I don't like the way you are approaching this.

Aren't you also assuming that if a player is not good in his rookie year, minute independent, it's on his own fault? Wouldn't a rookie be worse if the coach ignored him in practice?
Back to top
View user's profile Send private message
Ed Küpfer



Joined: 30 Dec 2004
Posts: 784
Location: Toronto

PostPosted: Wed Sep 28, 2005 9:58 pm Post subject: Reply with quote
mateo82 wrote:
How can you tell if a player is good unless they get quality minutes? You're assuming that a rookie who gets 15 mpg should be able to play up to his capabilities, as should a player who plays 30mpg or 45mpg.


That's not what I'm doing. I'm not worrying about whther the rookie is "good" or not—I'm using PER to see if he played well. The idea is, if a player plays well, he gets more playing time. If he doesn't, he gets less playing time. That is the way it should be, and it matches the data we have. On average, for every point of PER, a rookie plays 37 additional minutes.

mateo82 wrote:
Wouldn't it be easier to take a list of all of the rookies a coach has had and then manually pluck out the bad players (players who were bad on multiple stops)?


I have no idea why this would be any better. It seems to me that it would only complicate things. I certainly don't have the time to go through 3000+ rookie seasons to pick out the bad players.

mateo82 wrote:
Aren't you also assuming that if a player is not good in his rookie year, minute independent, it's on his own fault? Wouldn't a rookie be worse if the coach ignored him in practice?


I don't know anything about fault. My analysis only looked for factors that correlated with playing time in a player's rookie year. A rookie's PER had a significant correlation with his playing time. I'm not following your objections.
_________________
ed
Back to top
View user's profile Send private message Send e-mail
Kurt



Joined: 10 Jan 2005
Posts: 30
Location: Los Angeles

PostPosted: Thu Sep 29, 2005 12:43 pm Post subject: Reply with quote
Very interesting stuff. I find it hard to be a coincidence that the coaches that play rookies the least are considered some of the game's greats. Using PER as a way to balance the amount a player roughly "should" have played makes sense to me on a gut level. However, how many quality rookies has Phil Jackson or Larry Brown (for example) had to deal with? Same with Red in Boston. These were coaches of teams with largely established cores who had little use for rookies because of the team situation (plus drafting late, so not getting impact guys). You have the data in front of you and I'm going off the top of my head, but does PER really balance out the lack of use of rookies on these teams?

Well, with Larry we may learn a little more this season about not having an established team.
Back to top
View user's profile Send private message Visit poster's website AIM Address
mateo82



Joined: 06 Aug 2005
Posts: 211


PostPosted: Thu Sep 29, 2005 3:06 pm Post subject: Reply with quote
Ed Küpfer wrote:

I don't know anything about fault. My analysis only looked for factors that correlated with playing time in a player's rookie year. A rookie's PER had a significant correlation with his playing time. I'm not following your objections.


Because you don't know what the correlation is. Is it the PER that determines playing time or is it the playing time that determines PER? If a coach doesn't like rookies he doesn't pay attention to them in practice, he doesn't instruct them, and then they don't play well. Likewise the opposite could be true. The coach might give rookies a chance, but they just aren't very good, and so they have a low PER and get their playing time reduced. You just don't know what is the cause of the correlation. So the method of judging a player's worth to determine if a coach has something against rookie seems inheritably flawed, the way you are going about it.
Back to top
View user's profile Send private message
mateo82



Joined: 06 Aug 2005
Posts: 211


PostPosted: Thu Sep 29, 2005 3:12 pm Post subject: Reply with quote
Kurt wrote:
Very interesting stuff. I find it hard to be a coincidence that the coaches that play rookies the least are considered some of the game's greats. Using PER as a way to balance the amount a player roughly "should" have played makes sense to me on a gut level. However, how many quality rookies has Phil Jackson or Larry Brown (for example) had to deal with? Same with Red in Boston. These were coaches of teams with largely established cores who had little use for rookies because of the team situation (plus drafting late, so not getting impact guys). You have the data in front of you and I'm going off the top of my head, but does PER really balance out the lack of use of rookies on these teams?


I've looked at this question with Larry Brown, although I haven't tried to quantify it the way Ed Küpfer is doing. Almost all of his rookies get fewer than 20MPG. Most get far less. The only that got starter type minutes (from my memory) is David Robinson, who was a 24 year old rookie. He got Sean Elloit only 25mpg. He gave Larry Hughes only 19 MPG, who still managed 9ppg a solid stat for a rookie.
Back to top
View user's profile Send private message
Kevin Pelton
Site Admin


Joined: 30 Dec 2004
Posts: 976
Location: Seattle

PostPosted: Thu Sep 29, 2005 3:31 pm Post subject: Reply with quote
Kurt wrote:
You have the data in front of you and I'm going off the top of my head, but does PER really balance out the lack of use of rookies on these teams?

That's what the draft pick variable does though, doesn't it? Yes, those picks are often traded, but if a team is picking 25th, they're much more likely to have a set rotation already than a team picking fifth.

mateo82 wrote:
If a coach doesn't like rookies he doesn't pay attention to them in practice, he doesn't instruct them, and then they don't play well.

That doesn't sound like any practice I've ever attended.

Head coaches aren't really responsible for skills development anyway; that's basically the province of assistants. Ultimately, how much time a rookie spends developing his or her skills seems much more dependent on the player's own intrinsic motivation than any coaching decision; this work largely comes after practice.
Back to top
View user's profile Send private message Send e-mail Visit poster's website
hpanic7342



Joined: 16 Feb 2005
Posts: 201


PostPosted: Thu Sep 29, 2005 3:52 pm Post subject: Reply with quote
Ed, shouldn't you be controlling for the teams' records? I had just assumed you'd done this, until I took a look at the variables and didn't see it there. It seems most likely that Red didn't play rookies much because he had a host of all-stars at every position. Same with every other successful coach.
Back to top
View user's profile Send private message Send e-mail
kjb



Joined: 03 Jan 2005
Posts: 861
Location: Washington, DC

PostPosted: Thu Sep 29, 2005 4:55 pm Post subject: Reply with quote
I think this is a worthwhile study and a good approach. I do like hpanic's suggestion of including a variable for team winning %.

I agree with the other Kevin that the draft pick variable does address the issue of teams with established rotations. This also blends with the PER variable since players picked later in the draft are likely to have PER scores lower than players picked earlier.

mateo: What's the significance of Larry Brown's rookies getting fewer than 20 minutes per game? I think that for it to be a significant fact, there has to be an evaluation of what other coaches do with their rookies, and an evaluation of what kind of minutes Brown's rookies should have gotten during their rookie seasons. Of course, once you do all that, you'll have a study much like Ed's. Smile
Back to top
View user's profile Send private message AIM Address Yahoo Messenger
Ed Küpfer



Joined: 30 Dec 2004
Posts: 784
Location: Toronto

PostPosted: Thu Sep 29, 2005 8:25 pm Post subject: Reply with quote
mateo82 wrote:
Ed Küpfer wrote:

I don't know anything about fault. My analysis only looked for factors that correlated with playing time in a player's rookie year. A rookie's PER had a significant correlation with his playing time. I'm not following your objections.


Because you don't know what the correlation is.


Ah, but I do! The correlation coefficient between PER and minutes played in the sample of rookies is 0.35. Among non-rookies, it's higher, around 0.55.

mateo82 wrote:
Is it the PER that determines playing time or is it the playing time that determines PER?


That I don't know. Correlation is not causation, as the saying goes.

But think about it from a coach's perspective: if you have this rookie, about whom you know very little except that he's played poorly whenever you've given him playing time, are you going to increase his minutes and hope that he gets better? Or are you going to cut his minutes until he shows improvement? You seem to be arguing for the first option.
_________________
ed
Back to top
View user's profile Send private message Send e-mail
Ed Küpfer



Joined: 30 Dec 2004
Posts: 784
Location: Toronto

PostPosted: Thu Sep 29, 2005 8:30 pm Post subject: Reply with quote
Kurt wrote:
I find it hard to be a coincidence that the coaches that play rookies the least are considered some of the game's greats.


Bear in mind only one coach, Red Auerbach, had a negative effect on rookie minutes at a statistically siginficant level. The effects of Phil, Larry, and the rest are negative as well, but I am not confident that the effect represents anything other than random variation.
_________________
ed
Back to top
View user's profile Send private message Send e-mail
mateo82



Joined: 06 Aug 2005
Posts: 211


PostPosted: Sat Oct 01, 2005 9:03 pm Post subject: Reply with quote
admin wrote:

That doesn't sound like any practice I've ever attended.

Head coaches aren't really responsible for skills development anyway; that's basically the province of assistants. Ultimately, how much time a rookie spends developing his or her skills seems much more dependent on the player's own intrinsic motivation than any coaching decision; this work largely comes after practice.


That's great, but I didn't say anything about skill development.

In the practices you attended did the coach instruct the players as to how he likes his offense, defense to run? Did he go over some of the plays? Did he tell individual players what they are doing wrong? In other words did he do any coaching? If not, what did he do? I'm assuming these were Clippers practices then, correct?

mateo82



Joined: 06 Aug 2005
Posts: 211


PostPosted: Sat Oct 01, 2005 9:08 pm Post subject: Reply with quote
Ed Küpfer wrote:

mateo82 wrote:
Is it the PER that determines playing time or is it the playing time that determines PER?


That I don't know. Correlation is not causation, as the saying goes.


Exactly, which is why this method is going to be fruitless.

Quote:
But think about it from a coach's perspective: if you have this rookie, about whom you know very little except that he's played poorly whenever you've given him playing time, are you going to increase his minutes and hope that he gets better? Or are you going to cut his minutes until he shows improvement? You seem to be arguing for the first option.


I'm not arguing for any option. You've just described 1 possible scenario. It could also be that coaches, for what ever reason, only trust veterans and don't pay attention to rookies. They don't instruct them during practice and they only give them garbage-time minutes when players don't try hard. And since these players aren't getting as many rebounds or as many steals because play is casual and slow, they have a lower PER. And so your stats are tainted. This is also a possibility.

The simple fact is that a coach can affect his players PER. With that being the case, using PER to show that coaches do not discriminate against rookies is a logical begging the question. It's a good problem to try to quantify but this particular one doesn't seem to work.
Back to top
View user's profile Send private message
mateo82



Joined: 06 Aug 2005
Posts: 211


PostPosted: Sat Oct 01, 2005 9:12 pm Post subject: Reply with quote
WizardsKev wrote:
I think this is a worthwhile study and a good approach. I do like hpanic's suggestion of including a variable for team winning %.

I agree with the other Kevin that the draft pick variable does address the issue of teams with established rotations. This also blends with the PER variable since players picked later in the draft are likely to have PER scores lower than players picked earlier.

mateo: What's the significance of Larry Brown's rookies getting fewer than 20 minutes per game?


That's typically the minutes of a 7th or 8th man. That's the significance. Either Larry Brown has been ridiculously unlucky in his long coaching career and never has had a rookie that deserved starter minutes (aside from 24 year old David Robinson) or he just doesn't like them.

Quote:
I think that for it to be a significant fact, there has to be an evaluation of what other coaches do with their rookies, and an evaluation of what kind of minutes Brown's rookies should have gotten during their rookie seasons. Of course, once you do all that, you'll have a study much like Ed's. Smile


Well, rookies start in the NBA all the time. But yeah, I'd like to investigate this further.
Back to top
View user's profile Send private message
Ed Küpfer



Joined: 30 Dec 2004
Posts: 785
Location: Toronto

PostPosted: Sat Oct 01, 2005 9:51 pm Post subject: Reply with quote
mateo82 wrote:
And since these players aren't getting as many rebounds or as many steals because play is casual and slow, they have a lower PER.


PER is a pace-adjusted rate stat, like points per minute. Garbage time give players just as much opportunities to increase their PER as other minutes. More, in fact, since the level of opposition is lower. Once again, if a player cannot excel in spot-duty situations, what makes anyone think he'll improve with more responsibility?

mateo82 wrote:
And so your stats are tainted. This is also a possibility.


It is also a possibility that I will wake up in the morning to find Michelle Pfeiffer laying next to me—it's just not very likely. It would be a waste of time to contemplate a possibility with no evidence to support it. If you can produce evidence to support either possibility—the "low minute players have lower PER by virtue of playing fewer minutes" theory or the "genie granted my fondest wish" theory—then lets see it. Otherwise, you'll have to excuse me if I consider an accusation of "tainted" data to signal the end of a fruitful back-and-forth.
_________________
ed
Back to top
View user's profile Send private message Send e-mail
Ben



Joined: 13 Jan 2005
Posts: 264
Location: Iowa City

PostPosted: Sat Oct 01, 2005 11:25 pm Post subject: Reply with quote
Just saw this thread. Really great stuff, Ed.

I do think team winning percentage would be an important variable to include. Ideally, we would probably want PER of other players on the team at the same position.
Back to top
View user's profile Send private message
kjb



Joined: 03 Jan 2005
Posts: 864
Location: Washington, DC

PostPosted: Sun Oct 02, 2005 7:00 am Post subject: Reply with quote
Ben wrote:
Ideally, we would probably want PER of other players on the team at the same position.


This could be a really good variable to add. It then can maybe start getting at what alternatives the coach had. If we found that Larry Brown (for example) limited rookie minutes even when the rookie's PER was better than the guy playing ahead of him -- that may show some bias against rookies. There's still a hole because although Larry Hughes had a better PER than Eric Snow (using one example), the Sixers may have been better with Snow and Iverson on the floor than they were with Iverson and Hughes. But, I still think it'd be an interesting variable to add. (Easy for me to add it -- Ed's the one doing all the work here. Wink )
Back to top
View user's profile Send private message AIM Address Yahoo Messenger
mateo82



Joined: 06 Aug 2005
Posts: 211


PostPosted: Sun Oct 02, 2005 12:38 pm Post subject: Reply with quote
Ed Küpfer:

Your stat is predicated on the idea that a coach can not, either positively or negatively, affect the way his players play. And if this is not true, if the coach can help/hurt his players, then the entire thing is one long-winded begging the question.

When faced against undeniable facts it's best just to scrap what you're doing and start over. Don't be one of the stat-guys who ignore problems in favor of making sure the stat fits. Another approach is necessary in order to prevent this obvious begging the question.
Back to top
View user's profile Send private message
jambalaya



Joined: 30 Jan 2005
Posts: 282


PostPosted: Sun Oct 02, 2005 1:56 pm Post subject: Avg rookie PER increase/ minute increase ratio by coach Reply with quote
Ed Küpfer wrote:
The idea is, if a player plays well, he gets more playing time. If he doesn't, he gets less playing time. That is the way it should be, and it matches the data we have. On average, for every point of PER, a rookie plays 37 additional minutes.




I appreciate the new research Ed. Would you be willing to share the average PER increase/ minute increase ratio by individual coaches? Or even graphs of the curves to get a sense of how good you have to be to see a major increase in minutes with a particular coach?

As others have noted, rookie minute increase is significantly affected by the quality of the player in front of them and the degree of winning and neither player or the coach could really be "blamed" for that situation. Very good coaches with very good teams are likely to not need big minutes from rookies.

Though the focus of this thread is rookie production, more significant to me might be does the player and coach tap the player's potential relatively quickly and fully so say by year 3 you are getting 85%-95% of a player's ultimate productivity. Less than some figure, maybe 75-80% after three years and maybe someone was slow but who is ahead of you is going to explain a lot of that still and sorting out how to assign responsibility for slowness beyond that factor to the player and coach is a debatable topic but probably not conclusively "proveable" either way.

Addendum: "On average, for every point of PER, a rookie plays 37 additional minutes." (presumably for the season) Compared to a base PER rated player of what level? 0? (Or 10, 15, league average for these rookies? What level was that?)

It doesnt seem that elastic, about a half minute extra per game per 1 additional PER rating? Would a 2 1/2 minute a game increase for a 20 PER over a 15 PER mean that coaches are being stingy with extra minutes or that a 15 PER guy is already pretty fully utilized (given the other talent available ahead of him) and there isnt much room for him to get more.

One measure of rate of change doesnt seem enough in this case. Maybe average minutes per game received by PER level for rookies would be a useful chart to see, on average and perhaps with a line for rookies of each of the named coaches too, and also compared to a leaguewide regardless of years of experience PER/avg. minutes a game played line.

If you wanted to show the development / utilization curve over the years you could also do 2nd, 3rd and maybe 4th year player lines.

And you also compare the rookie year average to the average for the last month of the season. If rookies get most of the minutes they "deserve" by the end of the year I am less willing to criticize the coaches and in fact would congratulate them.

Along the way they give some of the limited available minutes to more veteran players because of the need for different skill mixes than provided by the rookie, specific difficult match ups, the need to keep everybody involved and sharp, to help the rookie avoid the wall, to give them someone to compete for minutes against, learn from, and to have an alternative in case the rookie chokes up down the season stretch or gets injured.

These additional research pieces would add lots of information to the debate but still leave plenty of room for different arguments about player /coach interaction, which really should be argued case by case, not on average and can't see "what might have been" if they played more that first year, a year where they are not exactly the same player as they will be down the road so those future stats arent "fully and simply applicable".

Last edited by jambalaya on Mon Oct 03, 2005 12:47 pm; edited 1 time in total
Back to top
View user's profile Send private message
Ed Küpfer



Joined: 30 Dec 2004
Posts: 785
Location: Toronto

PostPosted: Mon Oct 03, 2005 12:32 pm Post subject: Reply with quote
mateo82 wrote:
Your stat is predicated on the idea that a coach can not, either positively or negatively, affect the way his players play. And if this is not true, if the coach can help/hurt his players, then the entire thing is one long-winded begging the question.


No. My analysis is predicated on the idea that a player who plays well will be given greater opportunities in the future. The facts bear this out.

I see you've backed off your previous objection that PER is biased against low-minute players. Good. Your new objection: coaches affect player's play, so let's call the whole thing off. Unfortunately, you present just as much evidence to support this objection as you did for the previous one, ie none. But don't worry, I'll do the grunt work.

Forgetting for the moment that all player stats are affected by coaching, and that therefore if your objection were valid we wouldn't be able to do any work whatsoever—forgetting that, of course a coach can affect how well a player plays. No one would deny this. Missing is an estimate of how much effect coaches have. Here: I've calculated the PER for all players in my study separated into two categories: player's career PER while playing under the coach in question, and the player's career PER under all other coaches combined. The correlation coefficient is 0.58. To give you a sense of context, the correlation coefficient between a player's PER in even numbered years and odd numbered years is 0.61. The latter gives an idea of how much natural variation there is in player's year-to-year play. The 0.58 correlation is smaller, but the difference is not statistically significant.

What this means is that as far as the stats can tell, the identity of a coach has as much effect on a player's PER as whether the season is an even numbered year or odd. Obviously, a coach affects a player's production to a certain extent, but because so many other factors come into play the data become so noisy the effect is lost. I know there are ways of quantifying this effect, but it is enough for me to know that a coach's identity does not bias a player's PER, which is good enough for the ambitions of this study.

mateo82 wrote:
When faced against undeniable facts it's best just to scrap what you're doing and start over. Don't be one of the stat-guys who ignore problems in favor of making sure the stat fits. Another approach is necessary in order to prevent this obvious begging the question.


You really like hyperbole, don't you? "Undeniable facts." "Obvious" and "long-winded begging the question." "Tainted data." Pretty bold talk from someone who doesn't present any evidence in support, and whose objections seem to miss the point of what I was trying to do. In any case, I believe I've answered all your objections, to my satisfaction if not your's.
_________________
ed
Back to top
View user's profile Send private message Send e-mail
Ed Küpfer



Joined: 30 Dec 2004
Posts: 785
Location: Toronto

PostPosted: Mon Oct 03, 2005 12:41 pm Post subject: Re: Avg rookie PER increase/ minute increase ratio by coach Reply with quote
jambalaya wrote:
Would you be willing to share the average PER increase/ minute increase ratio by individual coaches?


The first post in this thread showed the results of the regression effects for the 25 coaches with most games coached. The second column estimates this effect—Red Auerbach's rookies, for example, played 349 minutes fewer than expected, on average.

But the important thing about that list is how few of coach's results were statistically significant: only 4 of the 25. If I ran the regression again, including more and more coaches, I would find fewer and fewer significant results, because the coaches I'd be adding would have fewer games coached. My personal interpretation of these results is that my methodology will not reveal the coaches—if there are any—with an undue reluctance towards playing rookies.
_________________
ed
Back to top
View user's profile Send private message Send e-mail
jambalaya



Joined: 30 Jan 2005
Posts: 282


PostPosted: Mon Oct 03, 2005 12:55 pm Post subject: Reply with quote
Right. I was just curious to see the findings displayed a different way. But the coaching effect does seem small, plus or minus about 5 minutes a game at most. It is their job to figure it all out, that seems like a still reasonable amount of room for the exercise of discretion with rookies.
Back to top
View user's profile Send private message
Ed Küpfer



Joined: 30 Dec 2004
Posts: 785
Location: Toronto

PostPosted: Mon Oct 03, 2005 6:49 pm Post subject: Reply with quote
Ben wrote:
I do think team winning percentage would be an important variable to include. Ideally, we would probably want PER of other players on the team at the same position.


I did a quickie regression on just the NBA teams from my original dataset, including this time team win%. You were right: it was very significant. One of the effects was essentially removing the Larry Brown's bias against rookies—after taking his team's win% into account, coach Brown is essentially neutral towards playing rookies. Huh.

I don't have time to run down the rest of the coaches. I'll probably get around to it in a few days.
_________________
ed
Back to top
View user's profile Send private message Send e-mail
Johnny Slick



Joined: 15 Oct 2005
Posts: 45
Location: Seattle

PostPosted: Tue Oct 18, 2005 2:56 am Post subject: Reply with quote
Ed Küpfer wrote:
Kurt wrote:
I find it hard to be a coincidence that the coaches that play rookies the least are considered some of the game's greats.


Bear in mind only one coach, Red Auerbach, had a negative effect on rookie minutes at a statistically siginficant level. The effects of Phil, Larry, and the rest are negative as well, but I am not confident that the effect represents anything other than random variation.
Yeah... although anecdotally speaking, that list makes a great deal of sense to me. George Karl's rep for not playing rookies is even greater than Brown's. His thing is in-game strategy, and you can't strategize with a player when you don't really know what his skill set is. You go with the guy who will get you 11 boards per 40 minutes over the guy who might give you 12 and might give you 8.

Back to more philosophical regression-related questions: it seems to me that guys like Karl and Brown are chosen, by and large, by veteran teams (or at least teams with few rookies) because of their established reputations. It's kind of hard to say that he's nearly as likely to play rooks as the next guy because he let Willie Anderson play a lot of minutes in San Antonio. I mean, he's had very few honestly bad teams in his career. That's a mark of his excellence as a coach but it's also a mark of his general refusal to work with rebuilding teams.
Back to top
View user's profile Send private message Yahoo Messenger
Ben



Joined: 13 Jan 2005
Posts: 264
Location: Iowa City

PostPosted: Tue Oct 18, 2005 1:57 pm Post subject: Reply with quote
Ed Küpfer wrote:
Ben wrote:
I do think team winning percentage would be an important variable to include. Ideally, we would probably want PER of other players on the team at the same position.


I did a quickie regression on just the NBA teams from my original dataset, including this time team win%. You were right: it was very significant. One of the effects was essentially removing the Larry Brown's bias against rookies—after taking his team's win% into account, coach Brown is essentially neutral towards playing rookies. Huh.

I don't have time to run down the rest of the coaches. I'll probably get around to it in a few days.


You might have been too busy to do this, but I just thought I'd let you know there's still interest in the results.

SGreenwell



Joined: 12 Feb 2005
Posts: 76
Location: Rhode Island

PostPosted: Fri Jul 29, 2005 11:16 pm Post subject: Coaches Exceeding Expected Winning Percentage? Reply with quote
Hi. I'm hoping I can get a fairly quick answer to this instead of having to search through three years of Prospectus'. I know Hollinger did a study in which he compared coaches expected winning percentage against their actual winning percentage and determined that JAckson, Brown, Saunders etc. had an ability to have their teams consistently perform better.

Is there a reference to this somewhere online, either in a similar study or the same one posted online? I'm looking to adapt this to the college game for an article I'm writing for a drafting web site.

Also, since I don't want to make two threads - Has there been work done determining which stats translate the best from the NCAAs to the pros? I recall someone saying that rebounds and blocks seemed to do the best, but again I'm looking for something published online. I want to study this issue a bit more, but I don't want to be reinventing the wheel either.

Final Q, I promise - Is there an easily accessible college basketball reference site, a la Basketball and Baseball Reference? Tracking down information for my work has been ridiculous so far, requiring me to comb through each conferences page.
Back to top
View user's profile Send private message Send e-mail Visit poster's website AIM Address
KnickerBlogger



Joined: 30 Dec 2004
Posts: 180


PostPosted: Sat Jul 30, 2005 12:46 am Post subject: Reply with quote
I don't know if this is what you're talking about, but I did it on a mini-scale with Wilkens:

http://www.knickerblogger.net/?p=25

Hard to believe he was 40-41 with New York.
_________________
KnickerBlogger.Net - now indispensable!
Back to top
View user's profile Send private message Visit poster's website AIM Address Yahoo Messenger
Kevin Pelton
Site Admin


Joined: 30 Dec 2004
Posts: 978
Location: Seattle

PostPosted: Sat Jul 30, 2005 12:59 am Post subject: Re: Coaches Exceeding Expected Winning Percentage? Reply with quote
SGreenwell wrote:
Is there a reference to this somewhere online, either in a similar study or the same one posted online? I'm looking to adapt this to the college game for an article I'm writing for a drafting web site.

I used Hollinger's method for the WNBA:
http://www.wnba.com/storm/news/donovan041201.html
Back to top
View user's profile Send private message Send e-mail Visit poster's website
SGreenwell



Joined: 12 Feb 2005
Posts: 76
Location: Rhode Island

PostPosted: Sat Jul 30, 2005 1:08 am Post subject: Reply with quote
Thanks, exactly what I was looking for.
Back to top
View user's profile Send private message Send e-mail Visit poster's website AIM Address
jkubatko



Joined: 05 Jan 2005
Posts: 702
Location: Columbus, OH

PostPosted: Sat Jul 30, 2005 3:48 pm Post subject: Reply with quote
This thread reminds me of something I was working on a while ago. In The Bill James Guide to Baseball Managers, Bill James outlines a system for rating baseball managers. I came up with a similar system for basketball. Coaches are awarded points using the following system:

1) +1 point if W > L
2) +1 point if W >= 60
3) +1 point if (W - L) >= 20
4) +1 point if division winner
5) +1 point if played in finals
6) +1 point if champion

Using this system, a "perfect" season is worth 6 points. Here are the all-time leaders:

Code:

Pat Riley 83
Red Auerbach 71
Larry Brown 70
Phil Jackson 64
Don Nelson 64
Lenny Wilkens 64
Jerry Sloan 56
Bill Fitch 50
Dick Motta 46
George Karl 44


Please note that I counted ABA and NBA seasons equally.

There are obviously many tweaks that could be made to this system, but I wanted to stick as close to James's system as possible.
_________________
Regards,
Justin Kubatko
Basketball-Reference.com
Back to top
View user's profile Send private message Send e-mail Visit poster's website
jkubatko



Joined: 05 Jan 2005
Posts: 702
Location: Columbus, OH

PostPosted: Sat Jul 30, 2005 4:03 pm Post subject: Reply with quote
I decided to look at wins versus expectation (WVE) as well. To make life easier I eliminated the first two seasons for all franchises, giving their coaches zero WVE. Here are all the all-time leaders:

Code:

Coach WVE WVE/82

Larry Brown 145.87 5.80
Don Nelson 91.32 3.62
Pat Riley 83.10 4.06
Phil Jackson 81.62 5.83
Red Auerbach 68.44 3.96
George Karl 67.83 4.48
Alex Hannum 60.97 4.12
Doug Collins 57.50 7.62
Bill Fitch 56.83 2.27
Lenny Wilkens 52.50 1.73


The last column (WVE/82) is WVE per 82 games.
_________________
Regards,
Justin Kubatko
Basketball-Reference.com
Back to top
View user's profile Send private message Send e-mail Visit poster's website
kjb



Joined: 03 Jan 2005
Posts: 864
Location: Washington, DC

PostPosted: Sat Jul 30, 2005 5:13 pm Post subject: Reply with quote
Justin: The career WVE -- is that column saying that over the course of his career, Larry Brown coached teams have won nearly 146 games MORE than the Pythagorean system would predict?
Back to top
View user's profile Send private message AIM Address Yahoo Messenger
HoopStudies



Joined: 30 Dec 2004
Posts: 705
Location: Near Philadelphia, PA

PostPosted: Sat Jul 30, 2005 5:30 pm Post subject: Re: Coaches Exceeding Expected Winning Percentage? Reply with quote
SGreenwell wrote:
Hi. I'm hoping I can get a fairly quick answer to this instead of having to search through three years of Prospectus'. I know Hollinger did a study in which he compared coaches expected winning percentage against their actual winning percentage and determined that JAckson, Brown, Saunders etc. had an ability to have their teams consistently perform better.

Is there a reference to this somewhere online, either in a similar study or the same one posted online? I'm looking to adapt this to the college game for an article I'm writing for a drafting web site.

Also, since I don't want to make two threads - Has there been work done determining which stats translate the best from the NCAAs to the pros? I recall someone saying that rebounds and blocks seemed to do the best, but again I'm looking for something published online. I want to study this issue a bit more, but I don't want to be reinventing the wheel either.

Final Q, I promise - Is there an easily accessible college basketball reference site, a la Basketball and Baseball Reference? Tracking down information for my work has been ridiculous so far, requiring me to comb through each conferences page.


This technique of comparing actual win-loss to pythagorean (or other type) win-loss projection is pretty heavily flawed. It doesn't get but a very small aspect of a coach's contribution. It also is highly dependent on the type of projection you use, something I show in BoP.

I would NEVER use it to evaluate a coach I was thinking of hiring.
_________________
Dean Oliver
Author, Basketball on Paper
The postings are my own & don't necess represent positions, strategies or opinions of employers.
Back to top
View user's profile Send private message Visit poster's website
jkubatko



Joined: 05 Jan 2005
Posts: 702
Location: Columbus, OH

PostPosted: Sat Jul 30, 2005 9:49 pm Post subject: Reply with quote
WizardsKev wrote:
Justin: The career WVE -- is that column saying that over the course of his career, Larry Brown coached teams have won nearly 146 games MORE than the Pythagorean system would predict?


Not compared to Pythagorean wins, no. Expected wins are calculated based on team winning percentage the previous season (weight 0.5), team winning percentage two seasons ago (weight 0.25), and a .500 winning percentage (weight 0.25). For example, suppose a team had a .600 winning percentage in 2005 and a .550 winning percentage in 2004. Their expected winning percentage for 2006 is .600*0.5 + .550*0.25 + .500*0.25 = .5625.
_________________
Regards,
Justin Kubatko
Basketball-Reference.com
Back to top
View user's profile Send private message Send e-mail Visit poster's website
jkubatko



Joined: 05 Jan 2005
Posts: 702
Location: Columbus, OH

PostPosted: Sat Jul 30, 2005 9:52 pm Post subject: Re: Coaches Exceeding Expected Winning Percentage? Reply with quote
HoopStudies wrote:
I would NEVER use it to evaluate a coach I was thinking of hiring.


I don't think anyone ever suggested such a thing.

I know your job requires you to focus on prospective analyses, but keep in mind that some of us prefer retrospective analyses. For that I think the methods I used above are reasonable.
_________________
Regards,
Justin Kubatko
Basketball-Reference.com
Back to top
View user's profile Send private message Send e-mail Visit poster's website
SGreenwell



Joined: 12 Feb 2005
Posts: 76
Location: Rhode Island

PostPosted: Sat Jul 30, 2005 10:00 pm Post subject: Reply with quote
Although I do have BoP, it is buried somewhere in two boxes of books that were also in my house last year, and I don't think I can unearth it this late in the day. Do the methods somehow favor coaches on better teams, or not seperate things well enough? I can't say that I fully remember the analysis of Billy Donovan as Florida's coach.
Back to top
View user's profile Send private message Send e-mail Visit poster's website AIM Address
HoopStudies



Joined: 30 Dec 2004
Posts: 705
Location: Near Philadelphia, PA

PostPosted: Sun Jul 31, 2005 11:20 am Post subject: Re: Coaches Exceeding Expected Winning Percentage? Reply with quote
jkubatko wrote:
HoopStudies wrote:
I would NEVER use it to evaluate a coach I was thinking of hiring.


I don't think anyone ever suggested such a thing.

I know your job requires you to focus on prospective analyses, but keep in mind that some of us prefer retrospective analyses. For that I think the methods I used above are reasonable.


I do like what you did more because that sets expectations based on prior seasons anc compares it to current season. I still wouldn't use it for hiring a coach for other reasons, but I think it sets a better base of info. But in-season win% v in-season pythagorean really just measures luck. How well do you win close games is, as Bill James said, mostly a measure of luck. I do think there is an element of skill in it, but very small. That being said, there are numerous economics papers that consider this a measure of manager skill (and I've made my argument to the economists). One of them went so far as to look for ability to learn in the data.
_________________
Dean Oliver
Author, Basketball on Paper
The postings are my own & don't necess represent positions, strategies or opinions of employers.
Back to top
View user's profile Send private message Visit poster's website
HoopStudies



Joined: 30 Dec 2004
Posts: 705
Location: Near Philadelphia, PA

PostPosted: Sun Jul 31, 2005 11:27 am Post subject: Reply with quote
SGreenwell wrote:
Although I do have BoP, it is buried somewhere in two boxes of books that were also in my house last year, and I don't think I can unearth it this late in the day. Do the methods somehow favor coaches on better teams, or not seperate things well enough? I can't say that I fully remember the analysis of Billy Donovan as Florida's coach.


The method of comparing reality to expectations favors coaches who keep good teams good and make bad teams better than would be expected. So, yes, coaches of good teams tend to make the top of the list. It's still a flawed approach, but a better framework... The reason I wouldn't use anything like this is because it doesn't suggest ways that a coach can get better other than "win more", which isn't advice.
_________________
Dean Oliver
Author, Basketball on Paper
The postings are my own & don't necess represent positions, strategies or opinions of employers.
Back to top
View user's profile Send private message Visit poster's website
jkubatko



Joined: 05 Jan 2005
Posts: 702
Location: Columbus, OH

PostPosted: Sun Jul 31, 2005 2:25 pm Post subject: Re: Coaches Exceeding Expected Winning Percentage? Reply with quote
HoopStudies wrote:
I do like what you did more because that sets expectations based on prior seasons anc compares it to current season.


Credit should really be given to Bill James (who came up with this method for baseball) and to John Hollinger (who applied it to basketball).

HoopStudies wrote:
I still wouldn't use it for hiring a coach for other reasons, but I think it sets a better base of info.


Once again, I don't think anyone ever suggested such a thing should be used to hire a coach. I'm not trying to answer the question "Which coach should I hire?" but rather "Which coach has accomplished the most?". Two very different questions. One question deals with forecasting the future, the other with evaluating the past.

HoopStudies wrote:
But in-season win% v in-season pythagorean really just measures luck. How well do you win close games is, as Bill James said, mostly a measure of luck.


I completely agree with the statement above.
_________________
Regards,
Justin Kubatko
Basketball-Reference.com
Back to top
View user's profile Send private message Send e-mail Visit poster's website
Dan Rosenbaum



Joined: 03 Jan 2005
Posts: 541
Location: Greensboro, North Carolina

PostPosted: Sun Jul 31, 2005 2:26 pm Post subject: Re: Coaches Exceeding Expected Winning Percentage? Reply with quote
HoopStudies wrote:
I do like what you did more because that sets expectations based on prior seasons anc compares it to current season. I still wouldn't use it for hiring a coach for other reasons, but I think it sets a better base of info. But in-season win% v in-season pythagorean really just measures luck. How well do you win close games is, as Bill James said, mostly a measure of luck. I do think there is an element of skill in it, but very small. That being said, there are numerous economics papers that consider this a measure of manager skill (and I've made my argument to the economists). One of them went so far as to look for ability to learn in the data.

In-season win% vs. in-season pythagorean is a good measure of which teams don't run up the score and/or which teams play worse in garbage time. That, of course, is a great way to measure quality coaching.

But seriously, I agree that comparing win% vs. some expectation based upon previous seasons is the best way to go. That is probably where I would start with some adjustments made for player changes and injuries - some adjustments that don't force Tim Floyd to have to win the championship with his Bulls team in order to make him look like a good coach.

page 2 0f 2 missing