APBRmetrics

Relationship of Selected Pre–NBA Career Variables to NBA Players’ Career Longevity

http://thesportjournal.org/article/rela ... -longevity



Mountain



Joined: 13 Mar 2007
Posts: 1527


PostPosted: Sat Sep 06, 2008 2:55 am Post subject: Rotations, depth, etc. Reply with quote
David Sparks had this recent contribution / post:

http://hardwoodparoxysm.blogspot.com/20 ... -team.html


A good idea can usually beget other derivative and / or creative ideas.

I initially thought that it would also be possible to calculate rotation depth to any quality standard perhaps declining as you go deeper or rotation continuity from previous year.

But you could also, I'd think, apply the Herfindahl Index to concentration of 5 man lineups- and that would be quite important too for better understanding what seems less understood that it should- lineups. And you could also look at concentration of winning 5 man lineups to gauge success of system or flexibility of players or coaching. Or identify how to attack the other guys pattern of success.

And look at winning distribution when a coach uses 8 or less, 8-9. 9-10 or 10+. Or high, average or low to team average or league average concentrations of 5 man lineups.

And look at these in relation to average team pace & winning or game pace or opponent average team pace and winning or opponent game H.I. for rotation or 5 man lineups or the net difference. Or other game factors.

Which works best when.

And I suppose you could look at player performance in these groups and subgroups and see who is "adaptable" or "consistent" or best within a certain usage pattern.

Slice, chop, sort, mix. So many things you could look at.

Last edited by Mountain on Sun Sep 14, 2008 1:30 am; edited 1 time in total
Back to top
View user's profile Send private message
Mountain



Joined: 13 Mar 2007
Posts: 1527


PostPosted: Sun Sep 14, 2008 12:02 am Post subject: Reply with quote
More on rotations and depth:

http://hardwoodparoxysm.blogspot.com/se ... rbitrarian

Wonder how the Finalists compare to league averages. I assume a bigger edge on the later but less clear if Finalists are elevated on the former.




As an aside if you were willing to share the minutes weighted team ages you calculated for this that would be of mild interest as an improvement upon the unweighted figures recently discussed.


And if you have done weighted team height by itself last season or against any performance measures for last season or thru time that would be of interest too.

Team height differential maybe be trivial but when it comes down to matchups height might sometimes matter though I haven't seen any comprehensive data compilation or analysis.

After noting the Lakers were tallest on unweighted last season I noticed Celtics were 8th shortest. Looking at past 6 years I see no pattern in Finalists for this and better numbers probably doesn't change that but it is curiosity especially if Boston - LA square off again with a healthy Bynum. presumably without PJ Brown and Posey gone and uncertainty about usable Boston wing depth and the height of it if Miles doesn't work out. Might matter more in the time bench players are on the court which isn't a lot. Of course it just one factor of many and far from most important. But will Odom or Radmanovic cause more problems from a SF spot?
Back to top
View user's profile Send private message
dsparks



Joined: 22 Feb 2008
Posts: 61


PostPosted: Sun Sep 14, 2008 2:31 pm Post subject: Reply with quote
Mountain: Here's a link to a google spreadsheet with several things calculated, for 614 team-years. It's got minutes-weighted age, and rotation and depth sizes, calculated at the entire season level (so they're not exactly rotation and depth, but more like concentration of contributions over the course of the season. Conc. ratio divides the concentration of MEV by the concentration of minutes. The idea here is that teams with smaller rotations will naturally have smaller depth, but there are some teams with little depth with relatively large rotations--the Concentration Ratio sort of controls for rotation size in calculating depth. So, the teams who rely massively on one or a few players, regardless of rotation size, are those with high CRs. Thus, you see some LeBron, Garnett and Jordan teams, and then some others with two players doing most everything (Utah, LAL), at the top.

On a whim, I divided team wins by the CR--I'm not sure exactly what it means--it's sort of like, "If the team produced in the same proportion as they shared floor time, this is how many wins they would have." Or perhaps, without the stars overproducing relative to playing time, how does the team fare? Or something like that. Those are "support wins" or "~star wins." Star wins is the difference between this number and actual wins.

So, I read star wins to indicate the relative contribution of the team's "best" players (in some sense), while support wins indicate the productivity of the team's "supporting cast." As you can tell, I still need some help interpreting just what these numbers could reasonably be said to mean (they might mean very little), but they're interesting regardless. According to the "support wins" measure, the strongest teams in terms of depth-productivity were BOS, HOU, DET, LAL...

http://spreadsheets.google.com/pub?key= ... 34cw&gid=0

Anyway, that's all I've got. I'd be interested to hear insights and interpretations.
_________________
David

http://arbitrarian.wordpress.com

Thanks for the documents with the average ages and this new information on star / support wins.

I see something of a relationship between what you've done and Dean Oliver's old article that he provided an update on. His article measured imbalance of possession usage and called it "Individuality Measure" . You measure imbalance of total contributions and called it "Concentration Ratio". Different but of related interest. Two levels of information, clearer than just eyeballing either set of raw data.

If I follow your description (thank you) correctly then it looks like the 2008 Rockets were the least dependent on stars for total contributions percentagewise or I think you can substitute "most balanced". This is praise for the support cast (and really all players) but is also a very modest report about the "extra" from the "Stars".

1993-94 Sonics got the most wins from their cast with little "extra" coming from the Star above the norm for the team based on minutes. I guess you could call them the most win productive cast of the very balanced teams.

Use of more possessions on offense is a big part of most Stars' extra contributions but it isn't the sole source available. Degree of efficiency in the use of those possessions obviously will affect your Concentration Ratio so the negative relationship that Dean finds for "Individuality Measure" and efficiency is an important part of understanding forces at work that affect your contribution numbers. But of course some Stars are more efficient than others and usage levels vary too. I guess it would be possible to split out the Star wins due to superior use of x offensive possessions vs other offensive and defensive contributions. If you did that it would somewhat similar to what I mentioned a year or two ago conceptually and labelled Player Factor Wins, an offshoot of a Team Factor Wins concept. What did a player do above league average on the 4 Factors that contributed to team wins.

Things can fit together.

Mountain



Joined: 13 Mar 2007
Posts: 800


PostPosted: Fri Sep 19, 2008 7:02 pm Post subject: Battle of the GMs Reply with quote
Win a title as GM and you obviously did some things right and probably had some luck.

I think there are 7 GMs who have been at the helm of multiple title winners since 1980. These are a higher level of made men, with a stronger case for being very good at their job.

2 of these are former NBA players, title winners as players for the Lakers- West and Kupchak.

While Kupchak and Buford try to add to their total, Dumars and Ainge - also title winners as players= look to get Title 2 (and maybe down the line Pfund / Riley) .

Others look to get their first.

Probably will be Ainge vs Kupchak again. Will they roll into playoffs looking as they do now or will one pull another trick out of their hat? Interesting roster adds, plenty of available trade chips. Should be an intriguing poker game & season.

Mountain



Joined: 13 Mar 2007
Posts: 800


PostPosted: Sat Sep 20, 2008 11:04 pm Post subject: Tom Thibodeau- Most Valuable Assistant Coach Reply with quote
It is well known that Boston was 1st on defensive efficiency and insiders know that Tom Thibodeau deserves a good share of the credit. Garnett anchored the D when he was on the court but that was less than 60% of regular season minutes. What might be worth highlighting is that when Garnett was off the court, under Thibodeau's (and Doc's) guidance, the Celtics points allowed per 48 was still good enough for 5th best in the league. Garnett made them #1 but Thibodeau clearly helped the rest of the players to raise their defensive performance.

Mike G



Joined: 14 Jan 2005
Posts: 2061
Location: Delphi, Indiana

PostPosted: Tue Sep 23, 2008 8:24 am Post subject: Ranking careers of recent entry classes Reply with quote
As a hobby, and mostly for fun, I've been ranking careers by statistics alone. Awards don't count, attitude is nothing; college and Olympics don't count. NBA playoffs count a lot. The whole (alltime) list is here:
http://spreadsheets.google.com/pub?key= ... KIuoUygRHw

Meanwhile, it looks as though we can size up the careers of several players who have been in the league 5 years or more. The vaunted entry class of '04 (rookies of 2003-04) look like this:
Code:
rank per36 rates G Min Eff% Sco Reb Ast PF Stl TO Blk T
45 Lebron James 437 42 .537 27.2 6.9 6.0 1.9 1.6 3.0 .7 40.4
70 Dwyane Wade 369 38 .554 26.4 5.1 6.3 2.7 1.7 3.8 .8 37.2
290 Carmelo Anthony 402 37 .533 24.9 6.3 2.8 3.1 1.1 3.1 .4 31.8

332 Chris Bosh 372 37 .554 20.4 9.3 2.0 2.7 .8 2.2 1.3 31.5
349 Josh Howard 400 32 .526 18.3 7.9 1.9 3.1 1.3 1.7 .7 28.3
391 Kirk Hinrich 410 35 .517 15.9 3.9 6.7 3.4 1.4 2.5 .3 26.5
These are 'standardized' rates, per-36-minutes and per-100 pts and 44 rebounds (per team, per game). Just arbitrary standards.

Five years ago, this was touted as the class of LeBron and Carmelo; by midseason, it was LeBron and Wade. The big gap is between Wade and 'Melo. Hinrich is close to Howard (rank) mostly by playing more minutes.

Here are 'equivalent totals', again combining seasons and playoffs. These are reconstituted from the per36 rates; as though they had been scoring/rebounding/etc in 100-pt, 44-reb games:
Code:
eT equivalents ePts eReb eAst ePF eStl eTO eBlk pom po/rs
19098 Lebron James 12858 3241 2825 896 744 1435 335 .108 .96
13930 Dwyane Wade 9873 1919 2340 998 632 1433 317 .154 1.00
12562 Carmelo Anthony 9816 2499 1115 1209 443 1241 175 .059 .80

11599 Chris Bosh 7515 3427 726 992 295 797 461 .023 .98
10142 Josh Howard 6544 2832 693 1121 450 618 251 .135 .91
10399 Kirk Hinrich 6226 1512 2627 1351 537 982 115 .049 1.07

'pom' are 'playoff minutes', showing the fraction of the player's minutes which are in playoffs. Bosh's 2.3% is very low. Hinrich and Anthony have been early outs.
'po/rs' is the player's career playoff T rate (see upper chart) divided by career season T rate. Wade's early thriving has been eroded in more recent showings. Carmelo has been a dog from the get-go.
_________________
`
There's no I in analysys.
Back to top
View user's profile Send private message Send e-mail
Mike G



Joined: 14 Jan 2005
Posts: 2061
Location: Delphi, Indiana

PostPosted: Tue Sep 23, 2008 1:00 pm Post subject: Reply with quote
The entry class of '03:
Code:
rank per36 rates G Min Eff% Sco Reb Ast PF Stl TO Blk T
87 Manu Ginobili 541 28 .582 23.3 5.7 4.7 3.1 1.9 2.7 .4 33.9
126 Amare Stoudemire 417 34 .594 26.1 9.9 1.3 3.8 1.0 2.8 1.7 36.6
161 Carlos Boozer 424 33 .566 20.9 12.6 2.6 3.4 1.0 2.5 .5 34.8
173 Yao Ming 423 33 .584 25.7 11.0 1.8 3.7 .4 3.0 2.1 37.4

276 Tayshaun Prince 566 34 .526 15.4 5.7 3.0 1.6 .7 1.4 .7 24.8
321 Mehmet Okur 527 28 .545 17.5 9.9 2.1 3.5 .6 1.8 .8 28.8
361 Caron Butler 445 36 .514 16.7 6.5 2.8 2.9 1.7 2.2 .3 25.9
614 Mike Dunleavy 481 30 .543 15.3 5.8 3.0 2.7 1.1 2.0 .4 23.6

Several of these guys have missed lots of games.

Equivalent totals, and playoff translations:
Code:
eT equivalents ePts eReb eAst eStl eTO eBlk pom po/rs
14916 Manu Ginobili 10252 2490 2076 828 1175 181 .226 .91
14267 Amare Stoudemire 10155 3864 495 392 1077 661 .090 1.04
13359 Carlos Boozer 8042 4838 993 400 946 206 .067 .95
14285 Yao Ming 9811 4199 678 164 1142 794 .047 .90

12928 Tayshaun Prince 8031 2960 1589 387 724 363 .234 .97
12267 Mehmet Okur 7470 4221 886 277 785 356 .113 .79
11258 Caron Butler 7243 2808 1225 753 956 133 .072 1.02
9548 Mike Dunleavy 6208 2370 1220 431 825 142 .000 .00

Ginobili has had 3 straight underachieving postseasons. Before that, he was great. He just had his best season and his worst playoffs.
Boozer ranks over Yao despite lesser numbers. His so-so playoff history is still much better than Yao's.
Manu and Prince, having been drafted into dynasties, are among the alltime leaders in % of minutes in playoffs (pom).
_________________
`
There's no I in analysys.

Last edited by Mike G on Tue Sep 23, 2008 2:39 pm; edited 2 times in total
Back to top
View user's profile Send private message Send e-mail
Mike G



Joined: 14 Jan 2005
Posts: 2061
Location: Delphi, Indiana

PostPosted: Tue Sep 23, 2008 1:09 pm Post subject: Reply with quote
Entry class of '02:
Code:
rank per36 rates G Min Eff% Sco Reb Ast PF Stl TO Blk T
133 Pau Gasol 536 36 .560 21.2 9.1 3.3 2.5 .5 2.6 1.9 33.8
135 Tony Parker 657 34 .532 20.6 3.7 6.3 2.2 1.1 2.8 .1 29.6
223 Gilbert Arenas 404 39 .550 24.0 4.3 5.3 3.1 1.7 3.3 .2 32.4
237 Richard Jefferson 567 36 .556 19.6 5.9 3.2 2.7 .9 2.3 .4 27.6

242 Andrei Kirilenko 530 31 .560 16.1 7.9 3.2 2.7 1.7 2.4 2.7 30.5
331 Zach Randolph 463 29 .514 20.6 11.0 1.9 2.9 .9 2.7 .3 31.1
355 Jason Richardson 531 36 .516 19.1 5.6 3.1 2.6 1.3 2.2 .5 27.6
411 Joe Johnson 566 36 .521 17.0 4.3 4.0 2.0 1.0 2.2 .3 24.8

530 Tyson Chandler 517 28 .576 11.7 12.2 1.1 4.1 .7 2.0 1.8 25.2
593 Shane Battier 583 33 .555 12.3 5.4 2.0 2.8 1.3 1.2 1.1 21.4
604 Troy Murphy 476 28 .520 14.2 10.1 1.7 3.5 .8 1.7 .5 25.3
621 Eddy Curry 501 26 .571 20.1 7.8 .8 4.1 .4 3.2 1.0 25.4

Equivalent totals:
Code:
eT equivalents ePts eReb eAst eStl eTO eBlk pom po/rs
17439 Pau Gasol 10954 4684 1699 278 1341 973 .059 .91
17981 Tony Parker 12512 2223 3800 656 1691 49 .192 .88
13532 Gilbert Arenas 9991 1803 2193 713 1383 100 .049 .93
15058 Richard Jefferson 10720 3228 1742 496 1260 209 .137 .91

14192 Andrei Kirilenko 7475 3661 1506 768 1117 1260 .080 .73
12009 Zach Randolph 7967 4234 740 330 1055 126 .018 .97
14213 Jason Richardson 9876 2876 1596 664 1136 255 .020 .90
13733 Joe Johnson 9434 2400 2215 546 1213 152 .038 .92

10380 Tyson Chandler 4821 5027 463 286 811 746 .043 .87
11496 Shane Battier 6572 2888 1088 685 629 587 .044 .71
9716 Troy Murphy 5463 3879 658 317 635 207 .000 .00
9618 Eddy Curry 7630 2975 290 161 1204 388 .000 .00

_________________
`
There's no I in analysys.

Last edited by Mike G on Tue Sep 23, 2008 2:40 pm; edited 1 time in total
Back to top
View user's profile Send private message Send e-mail
tiongkiat



Joined: 30 Apr 2007
Posts: 11


PostPosted: Tue Sep 23, 2008 1:56 pm Post subject: Reply with quote
Dunleavy was the 3rd overall pick behind Yao and Jay Williams in 2002, so he should be moved over to the class of '03.
Back to top
View user's profile Send private message
Mike G



Joined: 14 Jan 2005
Posts: 2061
Location: Delphi, Indiana

PostPosted: Tue Sep 23, 2008 2:42 pm Post subject: Reply with quote
Thanks, tiongkiat.
_________________
`
There's no I in analysys.
Back to top
View user's profile Send private message Send e-mail
tiongkiat



Joined: 30 Apr 2007
Posts: 11


PostPosted: Tue Sep 23, 2008 10:20 pm Post subject: Reply with quote
Thanks for posting the list. There are some interesting results that I wouldn't have guessed from my conventional wisdom.

- Nowitzki is #30 overall and his po/rs is 0.95, higher than I would have expected given all the not-clutch criticism he takes.
- Bryant has a po/rs of "only" 0.91. Which stretch of his career is most responsible for this ratio being so low, or has he consistently been a playoff underperformer?
- Marion po/rs of 0.86 (!). I guess I haven't watched him closely in the playoffs.
- R Wallace (0.94), Billups (1.03), Hamilton (1.03), Prince (0.97), B Wallace (1.04)/McDyess (0.90) as a unit look more consistent than Duncan (1.00), Ginobli (0.91), and Parker (0.88 ). Single playoff season performances (or luck or both) seem to matter more than consistency for championships.
- Isiah Thomas (1.10). I knew he was good in the playoffs, but having the highest ratio of your top 50 is a surprise.
Back to top
View user's profile Send private message
Mike G



Joined: 14 Jan 2005
Posts: 2061
Location: Delphi, Indiana

PostPosted: Wed Sep 24, 2008 5:06 am Post subject: Reply with quote
Good questions and good observations. There are actually two different things one could call 'po/rs' ratio. To rank a career, I like to see a 'playoff subcareer', and add that in. For this, a ratio of career playoffs to career RS works best.

But for a given postseason, over- and under-performers are generally judged relative to that season. Nowitzki played his first 2 years without making the playoffs. These were also his weakest years. For years in the playoffs, weighted by playoff minutes each year, his average playoff has been just .90 of his corresponding season.

This compares to an average of .94, so in fact is not good. Those Pistons you name have been pretty good. In Kobe's 5 trips to the Finals, he's been .84, 1.13, .91, .. .95 .. .89 last year. So yes, Shaq carried those Lakers, except for '01, when no one slowed them.

Superstars may be more likely to have their biggest playoffs following their biggest years. Role players are more random. Bill Cartwright was huge in NY, a dog with the Bulls. His 'annual' po/rs is a mediocre .89; for career, just .765 .

I've got career summaries for 295 current players. In order for their cumulative playoffs production over/unders to sum to zero, I have to divide their playoff rates by .9239 . This is considerably less than my off-the-cuff historic average of .94 . Lately, there's been an obvious dropoff in assists, come playoff time; so maybe that is reconcilable (see concurrent thread).

Here are the cumulative leaders, estimated, for their career playoffs, relative to that .9239 average; in playoff points added to their teams:
Code:
extra playoff guys po/rs annual Min
309 Tim Duncan 1.00 1.00 6204
231 Robert Horry 1.08 1.07 6825
170 Baron Davis 1.15 1.13 1754
137 Tayshaun Prince .97 1.02 4261
99 Caron Butler 1.02 1.17 1018

96 Shaquille O'Neal .96 .94 7658
82 Richard Hamilton 1.03 .97 4600
74 Dwyane Wade 1.00 .98 2202
72 Allen Iverson 1.01 .96 3205
71 Derek Fisher 1.01 1.00 4115

70 Tim Thomas 1.11 1.10 1527
68 Tracy McGrady 1.04 1.00 1604
59 Antonio Daniels 1.13 1.08 1793
51 Ben Wallace 1.04 .96 4277
51 Paul Pierce 1.00 .96 2547
'po/rs' is total career rates ratio
'annual' is the ratio that determines the 'extra' column. Shaq's .943 is still .019 better than the norm. Multiplied by his huge minutes and productivity, he still has a bit left to squander.

Cumulative underproducers:
Code:
extra playoff guys? po/rs annual Min
-85 Shawn Marion .86 .85 2246
-80 Tony Parker .88 .87 3733
-71 Peja Stojakovic .87 .84 2190
-61 Brent Barry .73 .74 1250
-60 Steve Nash .97 .88 3186

-60 Kobe Bryant .91 .90 5338
-57 Pau Gasol .91 .81 1026
-55 Dirk Nowitzki .95 .90 3290
-52 Jermaine O'Neal .94 .84 1694
-51 Michael Finley .85 .88 3185

-49 Mehmet Okur .79 .81 1322
-49 Carlos Boozer .95 .81 883
-47 Sam Cassell .82 .88 3127
-46 Jerry Stackhouse .82 .85 1695
-46 Carmelo Anthony .80 .79 686

The negatives are on a smaller scale than the positives above. These guys tend not to keep playing, and so have been early outs (notable exceptions).
_________________
`
There's no I in analysys.
Back to top
View user's profile Send private message Send e-mail
Mike G



Joined: 14 Jan 2005
Posts: 2061
Location: Delphi, Indiana

PostPosted: Wed Sep 24, 2008 6:08 am Post subject: Reply with quote
More observations:
The po/rs isn't directly figured into my alltime rankings. It says something about the size of a guy's playoff sub-career. But those playoff productions and rates are what counts.

Breaking the list into 100's, here are the average po/rs:
Code:
ranks po/rs
1 100 .977
101 200 .952
201 300 .944
301 400 .937
401 500 .931
501 600 .904
601 629 .896


The top 100 obviously consists of elite players, durable stalwarts and such. The bottom 129 tend to be more average players. Virtually none of the 629 are career scrubs (Chris Dudley?).

119 were active last year. Their average po/rs is only .912 . Again, an abundance of players who have been quickly out of the playoffs, and a relative few who get more minutes.

Which is the correct definition of average? The average NBA playoff player-minute? Or the performance of the average player, regardless of minutes?

Should a Nowitzki be held to the .98 standard of his 'ranking peers'? Or to the .90-ish level of the rank and file?
_________________
`
There's no I in analysys.
Back to top
View user's profile Send private message Send e-mail
Mike G



Joined: 14 Jan 2005
Posts: 2061
Location: Delphi, Indiana

PostPosted: Wed Sep 24, 2008 7:06 am Post subject: Reply with quote
Nowitzki should be held to a lower standard po/rs than, say, Ben Wallace.
Of players active last year, I find just over 1000 player-postseasons since 1999 among them. This is when the dominance of the Western Conference began.

Of 481 East player-postseasons, the mean is just .886. However, this includes many scrubby appearances, like Adonal Foyle's 3 minutes (in 3 games) last year. Total productions were .920 of the same-seasons, per minute, standardized to the best of my ability.

In 536 appearances by West players, only .897 of regular-season rates were met. So in the last 10 seasons, a West player can expect his playoff rates to drop .023 more than if he were in the East.

Together, the rate totals .909 po/rs. This is a bit less than the .912 reported above, due to a slightly different sample.
_________________
`
There's no I in analysys.

rlee



Joined: 20 Aug 2005
Posts: 39


PostPosted: Sun Sep 21, 2008 4:31 pm Post subject: "Why do Leaders Matter? The Role of Expert Knowledge&qu Reply with quote
Has any one here reviewed this?:

http://papers.ssrn.com/sol3/papers.cfm? ... id=1158980


From the summary in the Atlantic Monthly:

The authors calculated NBA coaches' winning percentages between 1996 and 2004 over 15,000 regular-season games. They found that coaches who had spent some time as a player got far better results out of their teams than those who hadn't.


The authors found that former all-stars tended to be better coaches than non-all stars. On average, teams with former all-stars as coaches placed six spots higher in league rankings than teams with coaches who had never played in the NBA, a huge bump-up in a league with only 29 total teams during the years studied.

One of their conclusions...

Leadership skills tend to derive from expert knowledge of a given trade, not from some mysterious alchemy of natural intelligence and interpersonal skills.
Back to top
View user's profile Send private message Send e-mail
Ryan J. Parker



Joined: 23 Mar 2007
Posts: 220
Location: Charleston, SC

PostPosted: Sun Sep 21, 2008 5:26 pm Post subject: Reply with quote
Thanks for the link. I'll see if I can make any sense of it. Very Happy
_________________
I am a basketball geek.
Back to top
View user's profile Send private message Visit poster's website
Mountain



Joined: 13 Mar 2007
Posts: 800


PostPosted: Mon Sep 22, 2008 2:36 am Post subject: Reply with quote
Gave it a quick read. Seems alirght as far it goes or as far as I as a non-stat expert can judge. But the time period is pretty short and the all-star coaches in the period generally had good teams (Sloan, Scott, Bird, Wikens, McHale, Westphal, Collins, Ainge) or average. Prior season records were checked and payroll effects addressed but the impact of roster turnover is not accounted for and presumably the all-star coaches generally selected good looking opportunities going forward (and knew when to bail too), while non-all-stars and non player coaches may have gotten more of the teams past their peak.
Back to top
View user's profile Send private message
mtamada



Joined: 28 Jan 2005
Posts: 200


PostPosted: Mon Sep 22, 2008 3:07 am Post subject: Reply with quote
Thanks for posting this Ray, I'd been meaning to post it myself, after seeing the same blurb in this month's issue of Atlantic magazine.

Here's my reaction after a very superficial reading of the actual paper: it looks like they did a decent job of specifying a model, and trying alternative specifications. It would take a closer reading and more thinking to decide if they left something out or had a major flaw but on the surface it looks like decent methodology.

My qualm with the paper is the time period covered. They say that they used NBA Guides and Registers from 1997 through 2004 (plus b-r.com and Rodney Fort's sports salary website). They don't explicitly state that those are the seasons they covered, but they do say that they covered 219 coach-seasons, so that's about 7 seasons, which presumably were within that 1997-2004 time frame.

I'd say that it was around the 1990s that we seemed to see the decline of the non-playing coaches (the Dick Mottas, Jack Ramsays, Bill Fitches, etc.) and a greater predominance of former players (Phil Jackson, Larry Brown, etc.). So the results may be valid for today's NBA, but I'm not sure if they would get the same results if they looked at say the 1970s and 1980s (of course, if one goes too far back, then one starts looking at coaches who lacked NBA experience because the NBA didn't even exist when they were of playing age).

And if that's correct, then it raises the question: what changed? Why is the "expert knowledge" more important now than it was then? One possible answer is that the NBA has become so much of a player's league that many players are prone to tune out the PJ Carlesimos but will still listen to a Nate McMillan. Dick Motta made a similar observation when he groused about how NBA coaches these days have to be "players' coaches" running simple schemes. (I'm reminded I need to write up and post about "My Breakfast with Dick Motta" that I had this past summer; I think he has a point with offensive schemes -- he complained that NBA offenses now consist of little more than repeated pick-and-roll plays -- but I should've asked him about defensive schemes, which strike me as more sophisticated now than they were 30 years ago.)

Another limitation of the study is the non-random nature of the coaches that they look at. I.e. most NBA all-stars never become NBA head coaches; the authors cannot say (and to their credit they do not claim) that NBA all-stars go on to become better coaches. Rather, among the people who have *already been selected to be NBA head coaches*, the former all-stars seem to do better.

Which is still an interesting observation, but of somewhat limited usefulness. You can't just go out and simply hire a former all-star to be your coach; you have to do the usual due diligence in selecting your coach, and maybe one candidate is a non-player but turns out to be Greg Popovich, and maybe one candidate is a former all-star but turns out to be, oh World B. Free or John Brisker.
Back to top
View user's profile Send private message
HoopStudies



Joined: 30 Dec 2004
Posts: 615
Location: Near Philadelphia, PA

PostPosted: Mon Sep 22, 2008 9:34 am Post subject: Reply with quote
I haven't read the article yet, but much of what MikeT mentions is on my list of things to look for when reading articles on coaching. Having done a lot on evaluating coaching, I have run into many difficulties. Certainly, generalizing to say that All-Stars or players are better than non-All-Stars or non-players is problematic for the reasons MikeT mentions.

But I will definitely take a look. Thanks for the link.
_________________
Dean Oliver
Author, Basketball on Paper
http://www.basketballonpaper.com
Back to top
View user's profile Send private message Visit poster's website
John Hollinger



Joined: 14 Feb 2005
Posts: 130


PostPosted: Tue Sep 23, 2008 6:25 am Post subject: Reply with quote
This post is worth it just for the John Brisker mention. Agreed with above, that you have to construct some type of win expectation for the team before you get into rating the coach. Otherwise the Tim Floyd vs. Phil Jackson comparison alone might account for much of the variation in the data.
Back to top
View user's profile Send private message AIM Address
Mike G



Joined: 14 Jan 2005
Posts: 2061
Location: Delphi, Indiana

PostPosted: Tue Sep 23, 2008 9:26 am Post subject: Reply with quote
Wasn't it just a couple of years back when 'common wisdom' held that former players do not tend to make good coaches?
I think I may have helped to debunk that notion.

Taking a select slice out of history, you can find just about whatever 'tendency' you want.
_________________
`
There's no I in analysys.
Back to top
View user's profile Send private message Send e-mail
mtamada



Joined: 28 Jan 2005
Posts: 200


PostPosted: Wed Sep 24, 2008 3:21 pm Post subject: Reply with quote
John Hollinger wrote:
This post is worth it just for the John Brisker mention. Agreed with above, that you have to construct some type of win expectation for the team before you get into rating the coach. Otherwise the Tim Floyd vs. Phil Jackson comparison alone might account for much of the variation in the data.


The authors mainly seemed to use payroll for create win expectations, which starts getting one to a ballpark figure, but obviously there's an issue of coaches who have to work for imcompetent GMs spending millions on a Jim McIlvaine vs coaches who work for ... not sure who the most efficient teams are in terms of wins per dollar (or marginal wins per marginal dollar to use a sabrmetric measure). I think they also used a fixed effects model (dummy variables for teams) to create separate modifiers for the win expectations for each team, i.e. to correct for teams which consistently do a good or poor job of spending salary on talent, this is all off the top of my head, haven't gone back to the article to check.

The major alternative of course is to use some sort of model of physical talent to create the win expectations, which is a consistent topic in this forum. Would that lead to better win expectations? Don't know, the trouble of course is that actual wins are determined by coaching as well as the team's talent level, and coaching is what they're trying to measure in the first place. And many of our models of player ability, to say nothing of player performance, can also be affected by coaches' use or mis-use of their players. (BTW Dick Motta had a quick answer for the question that we asked years ago on the old apbr-analysis email list: when I asked him about the sky-high assist numbers that centers such as Awtrey and Clifford Ray got when they played for his Bulls teams, he quickly said "It was the system": his offense looked to set up the forwards but if they didn't get an easy scoring opportunity then a lot of the offense flowed through the center. So Boerwinkle might look like one of the best passing centers in NBA history, but if he'd ever left the Bulls we might've seen him get fewer assists. I.e. we have to be aware of how coaching might affect our measures which claim to measure player ability.)

Yeah, in any history of the NBA or ABA, you could fill up half a chapter just talking about John Brisker.
Back to top
View user's profile Send private message
Harold Almonte



Joined: 04 Aug 2006
Posts: 501


PostPosted: Thu Sep 25, 2008 6:26 pm Post subject: Reply with quote
I wouldn't believe competent echonomists would use teams's payrolls to build a coach win expectation model, not after all the stuff talked about that kind of correlations last times.

We've seen some struggling with basic accounting, but everybody has done it with knowledge of that, although you wouldn't expect it from an echonomist.
Back to top
View user's profile Send private message
Mountain



Joined: 13 Mar 2007
Posts: 800


PostPosted: Fri Sep 26, 2008 8:21 am Post subject: Reply with quote
This study was about regular season wins but the current team quality inherited remains a bit uncertain.

But can't we move forward by looking at former All-Star vs normal player vs not a player coaches with comparable win levels in regular season and see how they do in the playoffs to get a somewhat better feel for coaching impact? What do those numbers show?


And exactly what do they do differently with regard to lineup composition, rotation, usage levels of stars vs support cast, offensive & defense efficiency or other stats (team shooting efficency, rebound differential, turnovers, fouls given, free throws, etc.), putting away wins or rallying from a big deficit, home and road winning %, breaking losing streaks or going on hot streaks?


Beyond that, in a simpler measure, I see 2 coaches who won titles as such since Billy Cunningham in 1983 who were All-Stars - 5 time All-Star Rudy Tomjanovich and 1 time All-Star Doc Rivers. 3 wins of the ultimate prize in 26 years isn't many though I don't know how many All-Star coach seasons that is compared to not and title win frequency difference.

I see 6 by non NBA players and 18 by non All-Star players.

The final test matters most.

It might be interesting to compare All-Star coach results with players who were on championship teams. Jackson won as a player, Riley and KC Jones. Am I missing somebody? That is I think 16 titles in 26 years.

Not sure that is really the magic reason (i.e. Jordan more important than Jackson) but it sure is a heck of a lot more impressive win total by these non All-Star player champions than the All-Stars.

All-Star GMs? I see Jerry West, Dumars and Ainge. So 1 multiple title winner as GM and just 2 others. Title winners as players I believe there are 4 (with 2 multiple winners).

Being an All-Star doesn't seem to a sufficient or perhaps even leading basis for being "a great leader" (judged by team results) as the vast majority of title winning coaches and GMs in last 26 years were not All-Stars. Being on a championship team as a player may indeed be important for winning a title as a Coach but the recent evidence is not unshakable proof.

2007-8 Rookie Rankings

Mike G



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

PostPosted: Sun Apr 13, 2008 3:12 pm Post subject: Reply with quote
Photo finish:
Code:
e82 per36 rates tm G Min Eff% Sco Reb Ast Stl TO Blk T
5.5 Horford,Al Atl 79 31 .535 12.5 12.2 1.5 .9 2.0 1.1 25.0
5.4 Durant,Kevin Sea 78 34 .507 19.8 4.5 1.9 1.0 3.1 1.0 23.9
5.1 Scola,Luis Hou 79 24 .537 17.4 10.2 1.7 1.1 1.9 .3 27.5
4.6 Moon,Jamario Tor 75 28 .541 12.2 8.7 1.3 1.3 .9 1.8 24.5
3.4 Young,Thaddeus Phi 72 21 .562 16.3 8.6 1.1 1.7 1.5 .2 25.6

3.4 Thornton,Al LAC 77 27 .496 15.9 6.0 1.4 .8 2.1 .7 21.4
3.0 Noah,Joakim Chi 71 20 .521 11.6 10.7 1.7 1.5 2.2 1.5 23.8
2.6 Landry,Carl Hou 39 17 .627 23.8 11.6 1.0 .9 1.2 .4 35.2
2.5 Williams,Sean NJ 72 18 .555 12.8 10.0 .7 .8 2.0 3.1 23.8
2.5 Navarro,JuanC Mem 80 26 .541 15.2 3.5 2.4 .8 2.2 .0 18.9

2.2 Jianlian,Yi Mil 66 25 .478 11.1 8.6 1.0 .8 2.0 1.2 19.5
2.0 Stuckey,Rodney Det 54 19 .472 14.6 4.8 5.3 1.7 2.6 .2 23.7
1.9 Dudley,Jared Cha 71 19 .521 11.0 7.6 1.8 1.4 1.3 .3 20.3
1.8 Green,Jeff Sea 78 28 .481 11.7 6.1 1.4 .7 2.6 .8 16.5

Yi doesn't even look like 2nd-team material.
_________________
`
36% of all statistics are wrong

Last edited by Mike G on Mon Apr 14, 2008 1:51 pm; edited 1 time in total
Back to top
View user's profile Send private message Send e-mail
S.K.



Joined: 18 Feb 2005
Posts: 61
Location: Toronto

PostPosted: Mon Apr 14, 2008 9:50 am Post subject: Reply with quote
Could we please start referring to either Horford or Thornton as "Allie" or some such thing? No matter which one is being referred to, I always think it's the other guy. I just spent ten minutes looking up stats to try to find out how on earth Thornton got himself to 12.2 reb/game.
_________________
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
tiongkiat



Joined: 30 Apr 2007
Posts: 11


PostPosted: Mon Apr 14, 2008 1:31 pm Post subject: Reply with quote
Only 2 PGs made the list (and pretty much at the bottom). How has that compared to previous years?
Back to top
View user's profile Send private message
Mike G



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

PostPosted: Mon Apr 14, 2008 2:18 pm Post subject: Reply with quote
Oh heck, somehow Chris Quinn got in there. He played last year (now edited out).

For the last 5 years, and just propagating the positions designated by my raw-data source (dougstats.com), here are rookie eWin totals by position.
Code:
year C PF SF SG PG rooks some PG's
2004 3.4 9.3 20.3 14.9 22.5 70.4 Wade, Hinrich, Lopez, Ford
2005 0.6 20.8 13.6 18.2 11.8 65.0 Nelson, Duhon, DHarris, Udrih
2006 11.4 12.0 17.5 3.9 31.4 76.2 Paul, Felton, DWilliams, Head,
2007 6.0 11.7 14.9 9.6 11.9 54.1 Foye, Rondo, MarWilliams
2008 13.9 14.0 23.1 3.3 7.8 62.1 Navarro, Stuckey, Conley


_________________
`
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: Tue Apr 15, 2008 10:26 pm Post subject: Reply with quote
Mike G wrote:
Photo finish:
Code:
e82 per36 rates tm G Min Eff% Sco Reb Ast Stl TO Blk T
5.5 Horford,Al Atl 79 31 .535 12.5 12.2 1.5 .9 2.0 1.1 25.0
5.4 Durant,Kevin Sea 78 34 .507 19.8 4.5 1.9 1.0 3.1 1.0 23.9
5.1 Scola,Luis Hou 79 24 .537 17.4 10.2 1.7 1.1 1.9 .3 27.5
4.6 Moon,Jamario Tor 75 28 .541 12.2 8.7 1.3 1.3 .9 1.8 24.5
3.4 Young,Thaddeus Phi 72 21 .562 16.3 8.6 1.1 1.7 1.5 .2 25.6

3.4 Thornton,Al LAC 77 27 .496 15.9 6.0 1.4 .8 2.1 .7 21.4
3.0 Noah,Joakim Chi 71 20 .521 11.6 10.7 1.7 1.5 2.2 1.5 23.8
2.6 Landry,Carl Hou 39 17 .627 23.8 11.6 1.0 .9 1.2 .4 35.2
2.5 Williams,Sean NJ 72 18 .555 12.8 10.0 .7 .8 2.0 3.1 23.8
2.5 Navarro,JuanC Mem 80 26 .541 15.2 3.5 2.4 .8 2.2 .0 18.9

2.2 Jianlian,Yi Mil 66 25 .478 11.1 8.6 1.0 .8 2.0 1.2 19.5
2.0 Stuckey,Rodney Det 54 19 .472 14.6 4.8 5.3 1.7 2.6 .2 23.7
1.9 Dudley,Jared Cha 71 19 .521 11.0 7.6 1.8 1.4 1.3 .3 20.3
1.8 Green,Jeff Sea 78 28 .481 11.7 6.1 1.4 .7 2.6 .8 16.5

Yi doesn't even look like 2nd-team material.


Just want to point out - Carl Landry was the 2nd highest player in the nation (to Durant) in my college player ratings last season. Nice to see him getting it done on the court in his very limited time.
_________________
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: Wed Apr 16, 2008 3:40 pm Post subject: Reply with quote
S.K. wrote:
Could we please start referring to either Horford or Thornton as "Allie" or some such thing? No matter which one is being referred to, I always think it's the other guy. ..

Bill Simmons asks, referring to Horford 'accidentally knocking T.J. Ford out of the lineup for six weeks', could we call him "A-Ho"?

http://sports.espn.go.com/espn/page2/st ... ons/080415
_________________
`
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: Fri May 02, 2008 8:40 am Post subject: Reply with quote
Durant received 90 first-place votes (545 points) from a panel of 125 writers and broadcasters. Atlanta’s Al Horford finished second with 390 points, and Houston’s Luis Scola was third with 146.

Durant wins by virtue of his final game: 42 points (18-25 FG), 13 reb, 6 ast. This was his highest-scoring game (by 5), most rebounds (by 4), and he only had more assists twice all year.

Horford's last 2 games were 8 pts apiece, total 6-19 FG. Both he and Scola got stronger as the season went along.

Code:
eWins per36 rates tm G Min Eff% Sco Reb Ast Stl TO Blk T
5.64 Durant,Kevin Sea 80 35 .507 19.9 4.6 1.9 1.0 3.1 1.0 25.1
5.46 Horford,Al Atl 81 31 .527 12.3 12.4 1.6 .9 2.0 1.1 25.8
5.22 Scola,Luis Hou 82 25 .533 17.7 10.3 1.8 1.1 1.9 .3 28.6
4.62 Moon,Jamario Tor 78 28 .533 12.1 8.9 1.3 1.4 .9 1.9 25.7
3.54 Young,Thaddeus Phi 74 21 .560 16.7 8.5 1.2 1.7 1.5 .2 26.6

3.38 Thornton,Al LAC 79 27 .491 15.8 6.1 1.4 .8 2.2 .7 22.0
3.21 Noah,Joakim Chi 74 21 .519 12.0 10.7 1.7 1.7 2.2 1.6 25.5
2.63 Williams,Sean NJ 73 18 .545 12.8 10.1 .7 .8 2.1 3.1 25.2
2.59 Landry,Carl Hou 42 17 .618 24.1 11.5 1.0 .9 1.3 .4 35.5
2.45 Navarro,Juan Mem 82 26 .531 15.0 3.6 2.5 .8 2.2 .0 19.4

2.15 Jianlian,Yi Mil 66 25 .474 11.1 8.7 1.0 .8 2.0 1.3 20.3
2.11 Stuckey,Rodney Det 57 19 .470 14.8 4.9 5.2 1.7 2.6 .2 24.6
1.98 Green,Jeff Sea 80 28 .480 11.8 6.2 1.5 .8 2.6 .8 17.6
1.94 Dudley,Jared Cha 73 19 .514 11.0 7.7 1.8 1.5 1.3 .3 21.0
1.68 Conley,Mike Mem 53 26 .491 11.9 3.8 4.8 1.2 2.4 .1 19.8

eWins per36 rates tm G Min Eff% Sco Reb Ast Stl TO Blk T
1.64 Young,Nick Was 75 15 .516 18.2 4.0 1.8 1.1 3.0 .2 21.1
1.47 Davis,Glen Bos 69 14 .521 14.0 9.0 1.1 1.2 2.5 .8 22.0
1.43 Smith,Jason Phi 76 15 .480 10.7 8.8 .6 .7 1.6 1.7 20.3
1.43 Hawes,Spencer Sac 71 13 .479 11.7 9.6 1.5 .6 2.3 1.6 21.9
1.37 Wright,Julian NO 57 11 .567 15.5 7.5 2.1 1.7 2.0 .7 25.8

1.34 Gray,Aaron Chi 61 10 .511 16.1 10.9 2.1 1.3 3.5 1.0 26.1
1.23 Brooks,Aaron Hou 51 12 .526 18.2 3.7 4.9 .8 2.7 .3 25.0
1.04 Afflalo,Arron Det 75 13 .481 10.8 5.9 1.9 1.2 1.3 .3 18.8
1.01 Cook,Daequan Mia 59 24 .476 11.9 4.6 1.6 .6 1.6 .2 16.5
.95 Brewer,Corey Min 79 23 .422 7.1 6.4 1.8 1.6 1.8 .5 15.3

.91 Chandler,Wilson NY 35 20 .471 12.0 7.3 1.5 .8 1.5 .9 20.5
.87 Sessions,Ramon Mil 17 26 .494 10.4 5.3 8.4 1.4 2.9 .2 24.5

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



Joined: 13 Mar 2007
Posts: 1527


PostPosted: Fri May 02, 2008 2:03 pm Post subject: Reply with quote
The list by eWins is based on actual minutes right?

For a different take I checked the T per 36 rankings

Landry,Carl
Scola,Luis
Young,Thaddeus
Gray,Aaron
Horford,Al
Wright,Julian
Moon,Jamario
Noah,Joakim
Williams,Sean
Durant,Kevin
Back to top
View user's profile Send private message
Mike G



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

PostPosted: Sun May 04, 2008 7:07 am Post subject: Reply with quote
Yes, eWins is generated by (T-R)*Min
where R is the T of a replacement-level player (who adds no wins).

The arguments for this award are many:
- Durant played for a bad team, got a lot of minutes, and was the team's offensive #1; drawing a lot of defensive attention from opponents. So his deficiencies are exaggerated.
- Scola played for an elite team, had to work his way into the rotation, didn't get a lot of minutes; so he would've gotten more attention (for better and for worse) on a weaker team.
- Horford's minutes, productivity, and team are in between those two.

Voters are more likely to go with per-minute numbers here. But also, a few years later, they probably expect Durant to be the better player; and this 'vindicates' a choice, it seems. Horford is 21, Scola 27, Durant 19.
_________________
`
36% of all statistics are wrong
Back to top
View user's profile Send private message Send e-mail
Mountain



Joined: 13 Mar 2007
Posts: 1527


PostPosted: Sun May 04, 2008 12:20 pm Post subject: Reply with quote
The ROYs for last 5 years played 30+ minutes a game so this year's choice only went 2 deep in main contenders.

Some of the high T rate guys with more modest initial playing time could see more growth in minutes than the top eWin in rookie season guys. We'll see who produces the best total product down the line.

Ultimately the wisdom of the draft picks will be measured in terms of player contribution to winning for a particular team.

First year is just a building block but many of the lottery picks could be described at disappointing so far. Including Green, Conley B Wright, Law, etc.

Is it more or less than normal?

Rookies playing 1200+ minutes at 14+ PER:

2003-4 5
2004-5 10
2005-6 7
2006-7 5
2007-8 6

I guess it is pretty normal for recent times.
.
Back to top
View user's profile Send private message
Mike G



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

PostPosted: Wed May 14, 2008 2:57 pm Post subject: Reply with quote
NEW YORK (AP)—Al Horford and Rookie of the Year Kevin Durant headlined the NBA All-Rookie team announced Tuesday.

Atlanta’s Horford was the only unanimous choice with 58 votes, followed by Seattle’s Durant with 57 in balloting by the league’s head coaches.

Houston’s Luis Scola (53), Los Angeles Clippers forward Al Thornton (48) and Seattle’s Jeff Green (43) also made the first team.
...

The All-Rookie second team was Toronto’s Jamario Moon, Memphis’ Juan Carlos Navarro, Philadelphia’s Thaddeus Young, Detroit’s Rodney Stuckey and Houston’s Carl Landry.

Jeff Green shouldn't even be 2nd team, should he?
No love for Joakim Noah, nor for Sean Williams.
_________________
`
36% of all statistics are wrong
Back to top
View user's profile Send private message Send e-mail
Mountain



Joined: 13 Mar 2007
Posts: 1527


PostPosted: Sun May 18, 2008 9:02 pm Post subject: Reply with quote
I took 82games' "net production" (1 on 1 matchup data) and basketballvalue's adjusted +/- and combined in similar fashion as the Roland Rating (2/3 weight on net production and 1/3 adjusted +/-) for roughly best half on own PER 1st rounders:

Player 1 on 1 Adjust +/- Adj. RR
Carl Landry, HOU 8.2 3.1 6.517
Jamario Moon, TOR 2.8 9.7 5.077
Thaddeus Young, PHI -0.8 12.3 3.523
Joakim Noah, CHI 0.9 5.6 2.451
Julian Wright, NOR 2.2 0.1 1.507
Luis Scola, HOU 0.3 1.6 0.729
Jared Dudley, CHA -3 2.1 -1.317
Kevin Durant, SEA 1.9 -8.8 -1.631
Juan CNavarro, MEM -1.8 -2.2 -1.932
Al Horford, ATL -2 -2.2 -2.066
Aaron Gray, CHI -6.4 3.5 -3.133
Mike Conley, MEM -3.8 -2.6 -3.404
Al Thornton, LAC -2.4 -5.6 -3.456
Glen Davis, BOS -3 -4.4 -3.462
Rodney Stuckey, DET -0.1 -10.4 -3.499
Sean Williams, NJN -2.8 -5.1 -3.559


Jeff Green, SEA -7.2 -6.6 -7.002


Landry and Moon move up to very top: Landry due to best 1 on 1 net production, Moon with 2nds best adjusted +/-. T. Young and Noah also benefit from strong adjusted +/-s (with Young the best). Horford and Durant slip to middle of this pack.

Aaron Gray worst on 1 on 1 data. Stuckey, Durant worst on adjusted +/-. (Jeff Green a low PER included as an exception to check on the #5 pick would be a runner=up for low marks on both.)

Mountain



Joined: 13 Mar 2007
Posts: 1492


PostPosted: Fri Mar 13, 2009 7:37 pm Post subject: Comments on some of Henry's 25 things learned at MiT list Reply with quote
11. Mark Cuban talked about a lot of the sophisticated statistics the Mavericks use. And then he was asked what kind of data they share with players. The example he gave was so basic ...-- they would tell someone if they were hot from a particular spot. Another team stat guy told me that he would keep his advanced insights to himself, unless it was encouragement to keep doing what they're doing, which was always welcomed by players and coaches. Mike Zarren says that on the Celtics, he shares what he believes to be important. "I've never been told: Don't tell me that," he reports. "I have," retorts the Nuggets' Dean Oliver. Oliver's case seems to be the more typical. Which makes me think that there must be a lot of insight that is not being put to use, and therefore there might be a little premium on players and coaches who are savvy in integrating this kind of input.


So who is going to be the first modern-era advanced analytics coach? Will it come from "the academy" / college or will it be a stats-savvy or taught to be stat savvy NBA player or assistant coach or I guess maybe some head coach could choose to be re-educated or might it be an analyst turned coach (or associate head coach?)?

Which schools train the players or coaches to be the most like advanced analytic guys or receptive to advanced analytic guys? How much is Battier just being Battier smart and how much it being the star student of Coach K.?

If Battier doesn't become Commissioner- probably not enough of a marketing or money focused guy, but maybe someday they have a true President of Basketball Operations in the real inner circle- would he make a good advance analytics GM? I don't think he is a coach type given his loner status he was dealt and choose.


19. Mark Cuban was clear that he thought a key factor in Kevin Garnett ending up in Boston was Danny Ainge's relationship with Minnesota's Kevin McHale. Cuban says the Mavericks were one of several teams that thought they had a deal done to acquire Garnett (he also mentioned Golden State) and had even heard from Garnett's agent who was wondering about an extension. But it didn't happen. "At the end of the day," says Cuban, with a wry smile, "relationships matter in the NBA."

Serendipity helped Ainge a lot but his ability to get what he wanted was a product of relationships- and maybe a product of estimates of him or his plan at the time. Folks don't usually hand an elite team what they wanted but a middling team at the time, that was easier to do, especially for a friend.

Ainge along with Riley and Bird and Thorn (did I miss anybody?) are the only GMs who have been an NBA player and coach. That gives a perspective advantage. I guess Ainge can communicate effectively with all levels of the game including the analysts.

For all the other aspects of being a GM how you fit into / work the GM phone tree is pretty important. I sense Ainge does that aspect of the game quite well.

I don't know how strong cooperative forces are in the league but there appear to be Boston. LA and San Antonio ex-colleagues who have significant instances of really helpful to the other cooperation not just perceived mutually advantageous deals. Not saying these are on-going networks but better to have good friends than not.

I doubt the league had a role in those trades or the Gasol trade but I wonder if the teams who helped those elite teams thought they would earn some goodwill from the league from it.



20. Mark Cuban: "Hiring coaches is the hardest job that there is. Period. End of story." Part of his assessment of Rick Carlisle, he explained, involved noticing that Carlisle very often played lineups that Cuban's database deemed to be the most effective from the available roster.

I tend to agree with that.


23. Mark Cuban says teams and the NBA suffer from "so many self-inflicted wounds." For instance, he couldn't fathom why the NBA would have a scheme where draft prospects could not be worked out in a five-on-five setting.

Yep. Yep. And could add more.
Back to top
View user's profile Send private message
BobboFitos



Joined: 21 Feb 2009
Posts: 8


PostPosted: Sat Mar 14, 2009 9:40 am Post subject: Reply with quote
Link to the full 25 list please?
_________________
-Rob
Back to top
View user's profile Send private message AIM Address
devin3807



Joined: 11 Oct 2007
Posts: 35


PostPosted: Sat Mar 14, 2009 11:53 am Post subject: Reply with quote
BobboFitos wrote:
Link to the full 25 list please?


http://myespn.go.com/blogs/truehoop/0-3 ... t-MIT.html

back2newbelf



Joined: 21 Jun 2005
Posts: 94


PostPosted: Sat Mar 14, 2009 9:43 am Post subject: A monkey with a laptop as coach Reply with quote
Hello

I had the idea of optimizing team lineups over an entire game, using raw lineup data like http://www.82games.com/0809/0809ATL2.HTM (broken down by possessions instead of minutes)

For every 4 minutes in a game I would (randomly) choose a certain lineup. This lineup would have a certain rating, depending on how good that lineup had been in games past.
I would sum the ratings of those (48/4=12) lineups, then look if the sum exceeds the prior best sum

I did put a couple of constraints in:
-no player can play more than 36 minutes
-a lineup has to have played together a minimum of X possessions so far in the season. (I chose X to be between 30 and 50)

the method has certain flaws:
-fouling is not taken into account
-it's not trying to match up good with the opponent lineup
-it's not an algorithm which comes up with the optimal solution, it just searches for a while and prints out the best it came up with


Why unadjusted lineup data?
I believe that, should a lineup be worse than their current (raw) rating, then their rating should drop in the minutes they get, giving them a lesser rating and thus lesser minutes in the next game.
Thus the raw ratings should adjust themselves (over a certain number of games)


Here is the method for this years' 76ers:
-all units with Elton Brand have been removed
-minimum # of possession played so far: 50
http://spreadsheets.google.com/pub?key= ... 7QYEHcSxfA

'rating' is how much that lineup is better than an average lineup in points/possession, an average lineup scores 1.07points/poss

an average of 0.14 over all used lineups would mean you should be outscoring an average opponent by 0.14points/poss, thus by 28 points in a game with 200 possessions(which is, obviously, too optimistic)

Last edited by back2newbelf on Sat Mar 14, 2009 2:34 pm; edited 1 time in total
Back to top
View user's profile Send private message
Mountain



Joined: 13 Mar 2007
Posts: 1492


PostPosted: Sat Mar 14, 2009 2:04 pm Post subject: Reply with quote
How good would a fan do armed with adjusted?

To simplfy to the extreme, what if you used these 3 lineups 36 minutes then filled in the rest with more study? (using Speights. a D league pure PG callup and whatever you have / fits best at wing)

Miller Green Iguodala Young Dalembert

Miller Williams Iguodala Young Ratliff

Miller Ivey Iguodala Young Dalembert

Essentially this is taking 3 lineups used for 40% of time and pumping them up based on adjusted and Brand's absence to 75%.

It is a numbers chase and you probably have to adjust based on results but I doubt it does worse than actual.

Compared to the Monkey I don't know. I don't give much credence to the ratings of low minute lineups but outside a few that is all the direct data you have to look at.

Last edited by Mountain on Sun Mar 15, 2009 2:05 am; edited 1 time in total
Back to top
View user's profile Send private message
back2newbelf



Joined: 21 Jun 2005
Posts: 94


PostPosted: Sat Mar 14, 2009 2:57 pm Post subject: Reply with quote
Mountain wrote:
How good would a fan do armed with adjusted?

What if you used these 3 lineups 36 minutes then filled in the rest with more study? (using Speights. a D league pure PG callup and whatever fits best at wing)

Miller Green Iguodala Young Dalembert

Miller Williams Iguodala Young Ratliff

Miller Ivey Iguodala Young Dalembert

you could potentially run into the problem that, for example, the 76ers do well enough with Iguodala and no Miller, and do well enough with Miller and no Iguodala, but be horrible without the both of them

consider
AB=+10
AX=+5
XB=+5
XX=-5
(A=Miller, B=Iguodala, X=whoever, you may use A and B exactly once)
Quote:

Compared to the Monkey I don't know. I don't give much credence to the ratings of low minute lineups but outside a few that is all you have to look at.

The lineups with low number of possession have that low of a number because it wasn't Monkey who coached them.
Those lineups that "look too good on raw +/- because their possession numbers are low" would have been used often, until their rating would have dropped to those of the other lineups.

Monkey would start the season trying all combinations (32 possible combinations with 10 players, 2 players for each position) for roughly 40 possessions, which should take around 6 games, then use the best ones. Throw in a unit with a bad rating once in a while, to see if they can improve their rating
Back to top
View user's profile Send private message
Mountain



Joined: 13 Mar 2007
Posts: 1492


PostPosted: Sat Mar 14, 2009 3:14 pm Post subject: Reply with quote
Totally free-form I'd trade as many of Miller, Williams, Green, Ivey as I could and re-do the guard staffing from scratch but I mainly stayed with what they had.

Starting open-minded playing lots of lineups early and getting more selective sounds good broad-brush. But if Monkey's opening set was 32 combinations then he is already the most selective coach in the league starting out. Most teams hit 100 pretty fast and get to 200+. I like the strategy of Monkey but injuries and other experiments probably raise the lineup set above 32 for most coaches.

Nearly every team was really scatter-shot last season and most players return so I'd try to sift and guess about those results and start with a fair amount of selectivity even early in the season (keep total lineups used say under 100 maybe even 75 or 50) and then get even more selective. That seems quite doable. The top 10 or 20 lineups should get a high percentage of the minutes and increasingly so.

I am not sure exactly how selective is the right answer, it depends on the team. But you & Monkey and I agree about not going too scattered and tightening with time.

Wonder how lineup concentration looks thru the monthly progression for real NBA coaches. Do they get more selective? If not, have they really learned much or are they just so confident in their ability to pick the right lineup for the moment? By raw and adjusted they appear to pick wrong a lot. It is a competition and someone loses while the other guy wins but after awhile I'd cut or cut off the weak performing lineups. Further than most NBA coaches do. Probably faster too but too fast could be an error as well of course and the main point is to eventually do it before the playoffs.

Mike G



Joined: 14 Jan 2005
Posts: 2363
Location: Delphi, Indiana

PostPosted: Sun Mar 15, 2009 10:46 am Post subject: Reply with quote
Lopez pulls away.
Gasol may overtake Mayo on the Grizz.
Westbrook isn't out of it.
Code:
e82 per36 rates tm Min Eff% Sco Reb Ast PF Stl TO Blk T
6.8 Lopez,Brook NJN 30 .556 16.1 10.6 1.0 3.9 .7 2.2 2.3 27.0
6.6 Rose,Derrick Chi 37 .499 16.0 3.9 5.3 1.5 .9 2.6 .2 24.2
6.5 Mayo,O.J. Mem 38 .529 18.9 4.5 2.5 2.5 1.1 2.7 .2 23.8
6.4 Gasol,Marc Mem 31 .575 15.1 10.2 1.8 3.8 .9 2.3 1.3 26.1
6.2 Westbrook,Russel Okl 33 .493 16.9 5.8 4.7 2.7 1.5 3.8 .2 24.9

5.6 Love,Kevin Min 25 .524 14.0 13.3 1.3 3.7 .6 2.0 .9 27.1
4.8 Beasley,Michael Mia 24 .510 18.7 7.8 1.1 3.2 .7 2.3 .7 25.6
4.6 Gordon,Eric LAC 33 .582 17.8 2.9 2.4 2.5 1.2 2.2 .5 22.1
4.2 Oden,Greg Por 23 .584 16.4 13.5 1.0 6.3 .8 2.5 1.9 29.2
4.1 Chalmers,Mario Mia 32 .543 12.6 3.3 4.9 3.5 2.3 2.4 .1 21.2

3.5 Thompson,Jason Sac 27 .519 12.7 9.6 1.2 5.0 .7 2.4 .8 21.3
3.4 Augustin,D.J. Cha 27 .566 16.7 2.6 4.3 2.6 .7 2.2 .0 22.4
3.3 Fernandez,Rudy Por 26 .579 15.1 4.2 2.6 1.8 1.0 1.6 .2 21.5
2.9 Mbah_A_Moute,Luc Mil 26 .515 9.9 9.3 1.1 3.7 1.5 1.8 .8 20.1
3.0 Speights,Marrees Phl 15 .555 16.7 8.1 .7 4.6 .9 1.3 1.6 25.5

e82 per36 rates tm Min Eff% Sco Reb Ast PF Stl TO Blk T
2.2 Batum,Nicolas Por 18 .536 11.6 6.7 1.8 3.6 1.4 1.4 1.1 20.7
2.1 Anderson,Ryan NJN 19 .528 12.4 8.9 1.2 4.3 1.0 1.7 .5 21.6
2.0 Hibbert,Roy Ind 13 .520 15.7 8.0 1.4 7.5 .5 2.0 2.7 24.5
1.9 Mcgee,Javale Was 15 .519 12.6 9.0 .4 4.3 .8 1.6 2.1 21.9
1.9 Arthur,Darrell Mem 20 .453 9.7 9.2 .8 5.2 1.2 1.3 1.2 19.9

1.8 Hill,George SAS 18 .503 12.2 4.5 3.4 3.8 1.1 1.9 .6 19.6
1.6 Randolph,Anthony GSW 15 .488 12.5 10.2 1.0 4.3 1.1 2.8 2.4 22.6
1.4 Lee,Courtney Orl 24 .556 12.1 3.3 1.6 2.8 1.4 1.4 .3 17.1
1.3 Morrow,Anthony GSW 19 .549 12.9 4.6 1.1 2.6 .9 .8 .2 18.7
1.2 Koufos,Kosta Uta 12 .532 12.3 8.3 .9 3.8 .7 1.4 1.7 21.4

_________________
`
There's no I in analysys.

Kevin Pelton
Site Admin


Joined: 30 Dec 2004
Posts: 828
Location: Seattle

PostPosted: Tue Mar 10, 2009 1:38 am Post subject: Morey-Simmons Podcast Reply with quote
http://sports.espn.go.com/espn/page2/simmons/index
http://sports.espn.go.com/espnradio/pod ... id=2864045

Fun listen. Apparently Michael Lewis has already agreed to headline next year's Sloan Conference. And clearly Simmons enjoyed his first trip to Dorkapalooza.

A noteworthy exchange:

Quote:
MOREY: Anyone who's out there saying, 'I've got this great system and it proves that this guy's better than that guy and anyone who thinks that guy is better than this guy is an idiot" -- that guy's an idiot. Anyone who's talking with any sort of level of confidence in their ability to know what's right, exactly how much someone is worth, they're the ones you've got to question.

SIMMONS: Unfortunately, I think those people kinda hurt the cause a little bit.

MOREY: They do. I think there's some folks out there like that and some that have gotten some attention and it has hurt the cause because they do talk with a level of confidence that isn't really warranted. No one should have that confidence, and especially those folks.
Back to top
View user's profile Send private message Send e-mail Visit poster's website
devin3807



Joined: 11 Oct 2007
Posts: 35


PostPosted: Tue Mar 10, 2009 10:47 am Post subject: Reply with quote
I haven't had a chance to listen to that yet (hopefully this weekend), but do you, or anyone else, think that statement is just sort of a general statement for anyone, or one geared at a specific person or persons?

I ask because one individual in particular comes to mind as being somewhat adamant in his system, while other experts have torn it apart. Further, according to The Hardball Times, at the conference, Cuban slammed this system as well.
Back to top
View user's profile Send private message AIM Address
davis21wylie2121



Joined: 13 Oct 2005
Posts: 693
Location: Atlanta, GA

PostPosted: Tue Mar 10, 2009 12:54 pm Post subject: Reply with quote
It was definitely a not-so-subtle bash of David Berri, because Simmons even specifically mentions this post where Berri writes, "...if we look beyond total points at the entire package, Brent [Barry] did produce more wins than his father. And that makes Brent the better Barry," as helping to discredit our statistical revolution. And you know what? I agree with Bill. A lot.
Back to top
View user's profile Send private message Visit poster's website
Westy



Joined: 15 Nov 2007
Posts: 14
Location: Chicago, IL

PostPosted: Tue Mar 10, 2009 3:24 pm Post subject: Reply with quote
Very good Podcast.

Good discussion of the conference, and gave further insight into the use of analytics in NBA operations.

Yeah, it was pretty obvious that Simmons doesn't think much of Berri's work, and it sounded like Morey concurs.
Back to top
View user's profile Send private message
Mountain



Joined: 13 Mar 2007
Posts: 1492


PostPosted: Fri Mar 13, 2009 10:02 pm Post subject: Reply with quote
i know it is not a simple issue and that folks feel Dave Berri over-credits individuals for their rebounding contribution which many think of as a team product (like defense in general) but I do see that defensive rebounding correlates- at team level- with actual and expected wins this season as 3rd best of the 8 parts of 4 factors offense and defense behind own FG% and FG% allowed. Berri also does put eFG% higher than some other systems and at least has a team defensive adjustment with a shot defense element whereas some other metrics have none, being agnostic from days when the data wasn't available but not changing from there either because it is still crude.
Back to top
View user's profile Send private message
Guy



Joined: 02 May 2007
Posts: 60


PostPosted: Sat Mar 14, 2009 1:53 pm Post subject: Reply with quote
Mountain: I don't think anyone disputes that DRebs have value at the team level. The question is whether WP is correct in equating each individual drb with a net additional team rebound. WP credits great rebounders with contributing 4 or 5 extra boards per game, or 300 to 400 per season.

But this cannot be true, because teams don't vary nearly that much in defensive rebounding. Last year, for example, the team SD for DRb% was just .016, which translates into about 50 rebounds. So a truly great rebounding team, +2 SD (98th percentile), will have about 100 Drb above average. That's valuable, but if individual players were truly adding hundreds of rebounds by themselves, we would see MUCH larger variation at the team level. If Berri were right, then every great rebounder in history has had teammates who -- coincidentally -- are FAR below average on rebounding (for their respective positions). Needless to say, this is implausible.

Eli and Cherokee have both done work demonstrating and measuring the large diminishing returns factor impacting Drbs.
Back to top
View user's profile Send private message
Mountain



Joined: 13 Mar 2007
Posts: 1492


PostPosted: Sat Mar 14, 2009 2:17 pm Post subject: Reply with quote
I agree. That is why I soft-pedaled the defensive rebounding comment. Some players add rebounds though. But I look for Aaron's adjusted rebounding data to estimate that, not raw rebounding which in WP rewards by role. PAWS position adjusts and might be a decent rough cut of impact for many / most player's impact. When we have adjusted rebounding we can see how well it does or doesn't do.
Back to top
View user's profile Send private message
magicmerl



Joined: 30 Dec 2007
Posts: 27


PostPosted: Sun Mar 15, 2009 10:26 pm Post subject: Reply with quote
For all the (justified) ripping of Wages of Wins, is it not fair to say that it's a better systems than the IMO vastly overhyped PER?
Back to top
View user's profile Send private message
davis21wylie2121



Joined: 13 Oct 2005
Posts: 693
Location: Atlanta, GA

PostPosted: Mon Mar 16, 2009 1:29 am Post subject: Reply with quote
magicmerl wrote:
For all the (justified) ripping of Wages of Wins, is it not fair to say that it's a better systems than the IMO vastly overhyped PER?

Two wrongs don't make a right...
Back to top
View user's profile Send private message Visit poster's website
Mike G



Joined: 14 Jan 2005
Posts: 2363
Location: Delphi, Indiana

PostPosted: Mon Mar 16, 2009 9:48 am Post subject: Reply with quote
It is not fair to say WoW is a better system than PER.
http://www.wagesofwins.com/AllPlayerMid0809.html

Gleaning the top (WP48) 24 on that page, and the top 24 PER's from
http://www.basketball-reference.com/fc/ ... i?id=ow1uN

We see that 12 guys are in both lists. Wade is 14th in Berri's list, 2nd in PER. Removing these 12, here are the remaining 12. Choose a side:
Code:
WP48 PER
.437 Camby 25.4 Kobe
.360 Kidd 24.2 Roy
.338 Przybilla 23.1 Jefferson
.332 Rondo 23.0 D Harris
.331 Biedrins 22.6 Nowitzki
.287 Murphy 22.4 Gasol
.281 McDyess 22.3 Shaq
.280 D Lee 22.1 T Parker
.268 Dampier 21.8 Bosh
.267 Calderon 21.4 Durant
.264 Okafor 20.9 Granger
.257 Bogut 20.6 Jamison
.251 Ariza 20.5 Amare
.248 C Andersen 20.5 D Williams

On the left are players PER doesn't list in its top 24, and WoW doesn't count Kobe, Dirk, et al among its top 24 : Chris Andersen is 'better' than all of them.
_________________
`
There's no I in analysys.
Back to top
View user's profile Send private message Send e-mail
bchaikin



Joined: 27 Jan 2005
Posts: 591
Location: cleveland, ohio

PostPosted: Mon Mar 16, 2009 11:09 am Post subject: Reply with quote
On the left are players PER doesn't list in its top 24, and WoW doesn't count Kobe, Dirk, et al among its top 24 : Chris Andersen is 'better' than all of them.

not taking sides here - but chris andersen is very good this year on a per minute basis. his stats extrapolated to 36 min/g and 82 games are 880 rebounds and 350 blocked shots. only 6 players have ever gotten 800+ rebounds with 300+ blocked shots in a season - jabbar, eaton, ewing, mutombo, olajuwon, and david robinson. as a matter of fact andersen's 08-09 stats, per touch, rebounding, and %BS, are very similar to those of mutombo from 95-96 (mutombo's best per minute shot blocking season), and those of alonzo mourning of 05-06 and 06-07 (with a similar high rate of fouls committed)...
Back to top
View user's profile Send private message Send e-mail Visit poster's website
Mountain



Joined: 13 Mar 2007
Posts: 1492


PostPosted: Mon Mar 16, 2009 1:20 pm Post subject: Reply with quote
Both systems say best at position

PG Paul
SG Wade
SF James
PF Duncan
C Howard

Second best?

WP PER

PG Kidd vs Harris
SG Kobe Kobe
SF Iggy vs Manu
PF David Lee vs Dirk
C Camby vs A Jefferson

Individual rebounding vs scoring

2 yr adjusted puts Garnett ahead of Howard on the top by position and Odom ahead of Duncan. Second best it has Kidd, R Fernandez, Iggy, Dirk and Nene.

I wouldn't follow any of these metrics completely.
But I'd guess they agree 65-80% on the top 5 by position.
Let folks argue about exact rank or the proper non-common fillers if they want.
Back to top
View user's profile Send private message
Harold Almonte



Joined: 04 Aug 2006
Posts: 576


PostPosted: Mon Mar 16, 2009 9:54 pm Post subject: Reply with quote
It's always healthy to remember the two WP metric's basic problems:

1- Not to mention that Rebounds take all the value of the shot defense (because its relative value obtained by regression); they are not rated. Players are not penalized by opposites's ORebs, and this is not included in the team defensive adjust either (not to mention that the off. rebounding action should have penalizations too). Total rebounds dominates half the metric.

2- The change of possession rule between teams after a FGMade, is taken as a bizarre way of individual punishment/prize distribution, leaving to high usage scorers the worst part. Of course this is an usage overpunishment in order to force individual level scoring break even, to team level scoring break even, and an "excesively minimalization" of scoring process, not to mention, an arbitrary and unrealistic approach about the usage-eff. tradeoff issue. This dominates the other half of the metric.

7SecondsorLess



Joined: 15 Feb 2009
Posts: 5


PostPosted: Sat Feb 21, 2009 1:15 pm Post subject: Who Are the New Shane Battiers? Reply with quote
Hey guys,

After reading Michael Lewis' article about how the Rockets viewed Battier as so underrated despite his not filling up the statsheet--that he just does the things that makes teams win--it got me thinking who the other really underrated guys are now that are maybe languishing on poor teams or possibly helping a great team but don't get nearly the credit they deserve.

Who would you say?

Off the top of my head, I would say Lamar Odom, Roger Mason Jr., and Dominic McGuire. Not great players, but Odom always gets bashed for not living up to his "full" potential--Shawn Marion heard much of the same with the Suns but he was still an outstanding player with them; I think Odom really does it all and I believe he has the best +/- on the Lakers this year. RMJ is just a sharpshooter who plays great D. I think he should get a lot more fame than he has for basically keeping the Spurs from any sort of deterioration this year; they're still just as good despite everyone saying they're too old and blah blah blah because he just comes in, plays good D, and drains 3's. Dominic McGuire clearly hasn't had the "winning" effect yet since the Wiz are tanking the season away but I think he's gonna be a stud role player down the line. He had 11 pts, 14 boards, and at least 3 blocks for the Wiz last night while completely shutting down Vince Carter; he is a lock-down defender who is an absolute monster on the glass. He doesn't have much offensive game yet but gets a ton of putback slams and can increasingly hit his shot from deep and keeps working on it.

If I'm very off on any of those guys, please tell me. I haven't read Basketball on Paper yet as I said in my intro thread and appreciated you all taking some time to educate me. I'm in the process of trading a friend for it and will be more conversant with you all in the future but just wanted to ask this question now. Thanks.
Back to top
View user's profile Send private message
Mountain



Joined: 13 Mar 2007
Posts: 1527


PostPosted: Sat Feb 21, 2009 2:59 pm Post subject: Reply with quote
The "new", contract season / challenged heavily by Jackson and Bryant Odom is versatile like Battier, showing the biggest positive adjusted +/- impact of his career. He has generally been a good adjusted guy but probably had his worst season in 2007-8. One or both of the twin pressure or being healthy seem to have worked.

I wouldn't put Mason and McGuire in the same category as Battier right now as they carry moderate and very negative adjusteds.

I'd call Jamario Moon a new Battier. Perhaps Thaddeus Young.

Different size but Turiaf and maybe Jason Thompson. Russell Westbrook doing much of it (not good shooting though) for a big positive adjusted with hustle, if not wisdom from years or precise discretion.

AK47 been doing it for years. Iggy does the make teammates better thing and his share of direct stuff.

Last edited by Mountain on Sat Feb 21, 2009 3:14 pm; edited 1 time in total
Back to top
View user's profile Send private message
Ed Küpfer



Joined: 30 Dec 2004
Posts: 787
Location: Toronto

PostPosted: Sat Feb 21, 2009 3:03 pm Post subject: Reply with quote
A side project I've never been able to get around to is the quantification of subjective opinions on player value. Once we have that, we can see who is "underrated" according to various player metrics.
_________________
ed
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: Sat Feb 21, 2009 3:11 pm Post subject: Reply with quote
Mountain wrote:
AK47 been doing it for years.

Yeah, in a sense Kirilenko is even more Battier-esque than Battier himself, just in terms of being a great defensive player and +/- darling who routinely gets disrespected by the conventional wisdom.
Back to top
View user's profile Send private message Visit poster's website
Mountain



Joined: 13 Mar 2007
Posts: 1527


PostPosted: Sat Feb 21, 2009 4:04 pm Post subject: Reply with quote
Ed Küpfer wrote:
A side project I've never been able to get around to is the quantification of subjective opinions on player value. Once we have that, we can see who is "underrated" according to various player metrics.


Would the All-Star vote list extended do?

Or if you are after "experts" would the the Annual GM survey work?
Back to top
View user's profile Send private message
thref23



Joined: 13 Aug 2007
Posts: 90


PostPosted: Sat Feb 21, 2009 5:49 pm Post subject: Reply with quote
I would add Amir Johnson.
Back to top
View user's profile Send private message Send e-mail
Kevin Pelton
Site Admin


Joined: 30 Dec 2004
Posts: 979
Location: Seattle

PostPosted: Sat Feb 21, 2009 6:10 pm Post subject: Reply with quote
davis21wylie2121 wrote:
Yeah, in a sense Kirilenko is even more Battier-esque than Battier himself, just in terms of being a great defensive player and +/- darling who routinely gets disrespected by the conventional wisdom.

True, although I'd note that Kirilenko's defensive value is more easily deduced from the box score because of the blocks and steals he accumulates, while Battier's defense is somewhat more difficult to quantify.

I'd throw out Chuck Hayes.
Back to top
View user's profile Send private message Send e-mail Visit poster's website
deepak



Joined: 26 Apr 2006
Posts: 665


PostPosted: Sat Feb 21, 2009 6:26 pm Post subject: Reply with quote
Kevin Pelton wrote:
davis21wylie2121 wrote:
Yeah, in a sense Kirilenko is even more Battier-esque than Battier himself, just in terms of being a great defensive player and +/- darling who routinely gets disrespected by the conventional wisdom.

True, although I'd note that Kirilenko's defensive value is more easily deduced from the box score because of the blocks and steals he accumulates, while Battier's defense is somewhat more difficult to quantify.

I'd throw out Chuck Hayes.


That's a good one. According to Houston Chronicle beat writer Jonathan Feigen, the Rockets staff rave about Hayes's defensive abilities. His role on the team has diminished quite a lot, though, as he hasn't fit in as well with the offense as he used to under Van Gundy and Landry has emerged as a much more effective and promising offensive player.

His strength defensively is he can switch on to forwards or centers on defense, and he's a very sound post defender. He's a smart team defender as well. I wonder how a player like that would fit on other teams, though. I don't think many teams would highly covet a player with seemingly no offensive skills, beyond chasing after the long offensive rebounds.
Back to top
View user's profile Send private message
7SecondsorLess



Joined: 15 Feb 2009
Posts: 5


PostPosted: Sat Feb 21, 2009 7:18 pm Post subject: Reply with quote
Kevin Pelton wrote:
davis21wylie2121 wrote:
Yeah, in a sense Kirilenko is even more Battier-esque than Battier himself, just in terms of being a great defensive player and +/- darling who routinely gets disrespected by the conventional wisdom.

True, although I'd note that Kirilenko's defensive value is more easily deduced from the box score because of the blocks and steals he accumulates, while Battier's defense is somewhat more difficult to quantify.


What about Battier's offense? I mean, I know it's very one-dimensional and he's never going to score a lot and doesn't have any moves or talent really but the article said that the Rockets' offense really flows when he's in and the ball always gets to the right spot while it's pretty stagnant without him. Why is that if he's not particularly talented? How can we find more players like that who aren't PGs who set up baskets well for teammates and seem to just play really well within the offense without doing much that shows up on the stat sheets?
Back to top
View user's profile Send private message
mtamada



Joined: 28 Jan 2005
Posts: 377


PostPosted: Mon Feb 23, 2009 7:22 pm Post subject: Re: Who Are the New Shane Battiers? Reply with quote
7SecondsorLess wrote:
RMJ is just a sharpshooter who plays great D.


To bring up a player from long ago, that description is an exact description of Dick Snyder (mainly played with the Sonics and Cavs). However his very high FG% made him not so invisible a contributor as Battier apparently is.

Snyder has been the best NBA player to come from Davidson College; Stephen Curry has a good chance but is not a shoe-in to overtake The Duck (I never did find out why that was his nickname; someone asked Bill Russell that question when he was coaching the Sonics and Russell's answer was "I dunno. Maybe he has webbed feet.")
Back to top
View user's profile Send private message
Mountain



Joined: 13 Mar 2007
Posts: 1527


PostPosted: Mon Feb 23, 2009 7:33 pm Post subject: Reply with quote
One can ask whether Amir Johnson still at a low 17 minutes a game is a statistical product of playing scrubs or being held back. Next season should answer. Always next season. Maybe Joe D doesn't want to prove the answer more fully ... or maybe he already has indirectly. 20+ PER first two season in extremely low minutes now down to sub 14 in still short minutes. Hard to say for sure in small minutes. Will have to play 25+ minutes to answer eventually.


(Stuckey aside: 4nd worst team offense and net differential with him on.)

Last edited by Mountain on Mon Feb 23, 2009 7:42 pm; edited 3 times in total
Back to top
View user's profile Send private message
deepak



Joined: 26 Apr 2006
Posts: 665


PostPosted: Mon Feb 23, 2009 7:34 pm Post subject: Reply with quote
In an online chat today on the Houston Chronicle website, this is how Shane Battier described his job:

Quote:

[Comment From KevinInSpring]
Shane, with T-Mac out for the rest of the year, how do you foresee your role changing on the court?

Shane Battier: I'll probably score a few more points, but for the most part, my role won't change much. My job is to get the ball to Yao, to spread the court, play great defense, get back in transition, and be a communicator. Those are all things that I do regardless of who is playing.


I just thought it was interesting the way he defined his role.
Back to top
View user's profile Send private message
thref23



Joined: 13 Aug 2007
Posts: 90


PostPosted: Mon Feb 23, 2009 11:33 pm Post subject: Reply with quote
Mountain wrote:
One can ask whether Amir Johnson still at a low 17 minutes a game is a statistical product of playing scrubs or being held back. Next season should answer. Always next season. Maybe Joe D doesn't want to prove the answer more fully ... or maybe he already has indirectly. 20+ PER first two season in extremely low minutes now down to sub 14 in still short minutes. Hard to say for sure in small minutes. Will have to play 25+ minutes to answer eventually.


(Stuckey aside: 4nd worst team offense and net differential with him on.)



The thing is, this season Amir has played a lot of his minute with the Pistons' starting unit, against other teams' starting units. His pure + / - (+6.3) is 2nd best on the team, and is just short of Rasheed Wallace (+6.5). Nobody else really comes too close. A great percentage of his minutes have come alongside Tayshaun Prince who has a negative plus minus. Maxiell McDyess and Kwame Brown - likely replacements when he doesn't start, are also a bit lower than him (although Kwame is a distant third in terms of + / -, at +1.0).

You could also point out that a large percentage of his minutes have come alongside Rasheed Wallace, and you could suggest that his positive +/- results from that....but you could say the same thing about Tayshaun Prince and he has a negative +/-
Back to top
View user's profile Send private message Send e-mail
Mountain



Joined: 13 Mar 2007
Posts: 1527


PostPosted: Tue Feb 24, 2009 1:10 am Post subject: Reply with quote
I shouldn't have said scrubs, that was in the past.

You are right he plays (or did) with Stuckey, Billups, Price and Wallace a lot this season. But probably mostly first and early third quarters?
Back to top
View user's profile Send private message
stareagle



Joined: 19 Feb 2009
Posts: 65


PostPosted: Tue Feb 24, 2009 1:25 am Post subject: Reply with quote
thref23 wrote:
The thing is, this season Amir has played a lot of his minute with the Pistons' starting unit, against other teams' starting units.


Yeah, Curry tends to either start him or give him a DNP-CD. Brilliant coaching there.

Being a stathead/sportswriter who happens to cover the Pistons, I can say that Amir is the one guy whose value I can not figure out. My own numbers have him as a very useful offensive player - he shoots for a high percentage, gets to the line a good amount and gets you some offensive rebounds.

But I also have him as a below-replacement-level defensive player, because of his ridiculous foul rate. He can not stay on the floor. As a result, he ranks as the second-worst player in Detroit's rotation. He's miles ahead of the horrific Arron Afflalo, and just behind the decaying corpse of Rip Hamilton.

(I've got McDyess first and Prince second, if you were wondering. No one else is even close.)

page 2 missing

stareagle



Joined: 19 Feb 2009
Posts: 12


PostPosted: Wed Feb 25, 2009 2:29 pm Post subject: Reply with quote
Mountain wrote:
By last season's defensive adjusted Johnson is a very good defender on average from a team perspective. From raw on/off and counterpart data at 82 games this season you get the same impression. Everyone get beats their share of times and maybe Amir looks bad on some of those plays but on average he's got good defensive numbers. I haven't seen him enough to rate him. But only 37% of his opponent's shots come inside which is quite low at PF. Maybe they should back him in more but they aren't. By contract 79% of his shots are inside. He's got a big edge on getting chip shots.


By my system, he was a below-average defender a year ago and is below-replacement this year, because of the fouls. I know that contradicts the +/- numbers, but my game-by-game stats show him as mediocre through the first week in January, and then falling off a cliff at about the same time as the team as a whole.
Back to top
View user's profile Send private message
Mountain



Joined: 13 Mar 2007
Posts: 1492


PostPosted: Wed Feb 25, 2009 3:18 pm Post subject: Reply with quote
The Pistons headed for the lottery?

I guess Amir Johnson is headed to sharing PF with Maxiell.
They'll need someone good to replace Wallace (and MCDyess).



Stareagle what do your numbers show on Stuckey's defense? His defense at PG looks weak by counterpart and team data to me. Team eFG% allowed on looks good but fouls go up and forced turnovers down.

He is a SG trying to a be a PG. Dumars could do it but not his protege, at least not yet. I don't sense he is ready to be a quality team starting PG / leader and is just a guy pretty much able to play individual ball though surely he tries. 2 yr Adjusted has him as 50th worst in league on impact. That is up from what it was, so I guess there is some learning and reason for hope but how much?

A fair number of surprises on that list. http://tinyurl.com/5apkgr


What is your sense of him overall?

And the prospects for Joe Dumar's new edition? Probably too early to say. But looks like they'll need major free agent scores and a new coach to get back on track.
Back to top
View user's profile Send private message
stareagle



Joined: 19 Feb 2009
Posts: 12


PostPosted: Wed Feb 25, 2009 3:57 pm Post subject: Reply with quote
Mountain wrote:
Stareagle what do your numbers show on Stuckey's defense? His defense at PG looks weak by counterpart and team data to me. Team eFG% allowed on looks good but fouls go up and forced turnovers down.

What is your sense of him overall?

And the prospects for Joe Dumar's new edition?


I've got Stuckey rated defensively as one of the worst starting PGs in the league. He was a little below-average last year, but has dropped off significantly. With the exception of Prince, every Pistons regular has fallen off some, but Hamilton, Stuckey and Johnson have been the worst offenders.

Overall, I think Stuckey has considerable talent - he did put up 38 and 40-point games in a span of 10 days - but has completely lost his confidence.

There was a stretch from about Thanksgiving through Christmas where he might have been their best player, and he was getting about 38% of their touches when he was on the floor. During the current losing streak, he's been a replacement-level player, and he's only getting about 18% of their touches, which is ridiculous for a point guard.

I think they have a decent chance to bounce back next year, but it will be a real test of Dumars' talent. He can't blow another lottery pick, and he's going to have to do something big with the cap space he's going to have. Then find a coach that can figure out what to do with the talent he has, instead of flailing around with a new plan every week.

Next year, Johnson is still only going to be 22 and Stuckey will be 24. Hamilton will be 31, but Maxiell and Afflalo are young. Get them a coach that can develop players, and they could be right back where they should be this year - battling for the last home-court spot in the East.
Back to top
View user's profile Send private message
back2newbelf



Joined: 21 Jun 2005
Posts: 94


PostPosted: Mon Mar 16, 2009 9:02 am Post subject: Reply with quote
I'd like to add Nicolas Batum and Luc Mbah a Moute to the discussion. Both players don't score alot of points but have a positive adjusted +/- rating, and in the case of Batum it's quite high.
Also, both are part of a top 5-man-unit
Back to top
View user's profile Send private message
Mountain



Joined: 13 Mar 2007
Posts: 1492


PostPosted: Mon Mar 16, 2009 2:45 pm Post subject: Reply with quote
Looking at raw on/off as a guide til the adjusted splits are available it appears that Batum's impact is on the offensive side of things. Team defensive efficiency is actuallly a bit worse when he is on despite the rep but as I said don't know the adjusted yet.

Luc appears from raw on/off to have a defensive impact but offset but weaker team offense.

Ryan J. Parker



Joined: 23 Mar 2007
Posts: 396
Location: Charleston, SC

PostPosted: Tue Mar 17, 2009 12:42 am Post subject: Putting context around rebounding rates Reply with quote
I'm trying to focus on putting useful context around important stats like FG%, reb%, etc, and this is my first attempt: effective rebounding rates.

What I'd like to know is, what context do you feel is useful? My short list includes lead/deficit, quarter/time remaining, and shot clock I must mention 82games.com, as Roland's work has provided me with a lot of good ideas for adding context.

That said, I'd like to hear everyone's thoughts on how we might add more context to the important stats like FG%, reb%, etc.
_________________
I am a basketball geek.
Back to top
View user's profile Send private message Visit poster's website
Mountain



Joined: 13 Mar 2007
Posts: 1492


PostPosted: Tue Mar 17, 2009 4:24 am Post subject: Reply with quote
You could argue that the better the opponent scores (from field or line) or protect the ball, the more important defensive rebounding is.

And on offense the worse your team is at scoring or turnovers the more important offensive rebounding is.

And more important when 3 pt frequency is elevated.
Blazers and Lakers only a bit above average on 3 pt frequency but very high on Offensive rebounding. Dallas a bit different but a pretty good combo. Cavs even better and top 10 in both. Celtics almost. Magic, Hornets and Spurs high on 3 pt attempts but very low on offensive rebounding. Rockets very high on offensive rebounding but a bit below average on 3 pt frequency (not pace adjusted). Utah very low on 3 pt frequency but very high on offensive rebounding.
Back to top
View user's profile Send private message
Mike G



Joined: 14 Jan 2005
Posts: 2363
Location: Delphi, Indiana

PostPosted: Tue Mar 17, 2009 4:59 am Post subject: Reply with quote
Reb% is supposed to be the % of available rebounds a player gets, right?
Since one's teammates are getting rebounds, those should probably not be considered 'available'.
Rather than Reb% = Reb/(TmRb + OpRb) -- which assumes equal 'availability' of all rebounds -- perhaps the formula should be:

Reb% = Reb/(TmRb + OpRb) * (TmRb/OpRb)^x

I've found an exponent ~ .50 works well.
_________________
`
There's no I in analysys.
Back to top
View user's profile Send private message Send e-mail
BobboFitos



Joined: 21 Feb 2009
Posts: 8


PostPosted: Tue Mar 17, 2009 5:43 am Post subject: Reply with quote
Really it shouldn't be %of rebounds available, but rather to the extent you're suppressing opponent OReb%. One thing that always makes me laugh is when teammates fight over rebounds - in the stats one player "gets" the board but in essence it really doesn't matter.

Opponent OReb% (on a team level, which can be broken down from the 5man to the individual) is probably the measure that should be looked at more critically.
_________________
-Rob
Back to top
View user's profile Send private message AIM Address
bastillon



Joined: 04 Nov 2008
Posts: 34


PostPosted: Tue Mar 17, 2009 9:32 am Post subject: Reply with quote
I don't use any specific formula, but if you want to come up with eRR% just adjust player's REB% to team's Rebounding Rate. The better your team rebounds, the higher your team is at Rebounding Rate, the less rebounds you'll grab. But there would have to be some research about correlation between them before one could make a stat like that.
Back to top
View user's profile Send private message Send e-mail
deepak_e



Joined: 26 Apr 2006
Posts: 455


PostPosted: Tue Mar 17, 2009 11:00 am Post subject: Reply with quote
I think Daryl Morey made the comment in that NYT Battier piece that rebounding isn't about how many rebounds you get, or even the percentage of overall rebounds you get, but mostly about how good you are at getting rebounds in your zone.

A player who's job is to play mostly the perimeter for floor spacing and be the first one back in transition may not get a lot of rebounds because he's not expected to be in position to get those rebounds. Should he be penalized for that? Tracking rebounding at that level would require more than the box scores or play by plays.

One thing that would be useful to distinguish is offensive rebounds off your own miss. Some players will tip the ball multiple times around the basket, and perhaps that inflates their rebounding ability on that end.

Last edited by deepak_e on Tue Mar 17, 2009 11:01 am; edited 1 time in total
Back to top
View user's profile Send private message
fundamentallysound



Joined: 18 Jul 2008
Posts: 14
Location: VA

PostPosted: Tue Mar 17, 2009 11:42 am Post subject: Reply with quote
BobboFitos wrote:

Opponent OReb% (on a team level, which can be broken down from the 5man to the individual) is probably the measure that should be looked at more critically.


Are you suggesting using Opponent Team OReb% and then running the adjusted plus minus methodology (like Ryan has been doing with shot selection, efg%, etc.) on that stat to determine on an individual level what the player's impact on Opponent Team OReb% is? That would be something that I think would be very valuable and I'd be interested in seeing. You could, of course, also do the converse and run the same methodology for differences in Opponent Team DReb% to see who are the best at suppressing the other teams ability to grab defensive rebounds (which should mean more offensive boards for their team).
Back to top
View user's profile Send private message AIM Address
deepak_e



Joined: 26 Apr 2006
Posts: 455


PostPosted: Tue Mar 17, 2009 11:55 am Post subject: Reply with quote
We could perhaps visualize a player's rebounding prowess as a rebounding curve, where the Y-axis is likelihood of getting the rebound, and X-axis is how far away the player needs to move from where he's standing when the shot is taken to get the rebound. Some players, like Yao Ming, may be very good at getting rebounds in their immediate area, but very poor at moving to get the rebound. Others, like Chuck Hayes, have better mobility for chasing down rebounds, but may aren't as good at securing it in their immediate area because of their lack of height.

Just a thought.
Back to top
View user's profile Send private message
Harold Almonte



Joined: 04 Aug 2006
Posts: 576


PostPosted: Tue Mar 17, 2009 5:00 pm Post subject: Reply with quote
It's obvious everybody knows that "availability of rebounds" is not the same at every position, and some good rebounders have a wider rebounding zone than others at same position. If we account for that (and for the same position opponents's ORs), a position adjust would make reb% with very low variations among positions.

Of course, doing that, we would be cancelling the rebounding usage issue. A mistake that shouldn't be committed like some metrics with scoring. Height and attached positions are skills, and the right balance between usage and eff. must be found in this basketball action, like in any other one.
Back to top
View user's profile Send private message
BobboFitos



Joined: 21 Feb 2009
Posts: 8


PostPosted: Tue Mar 17, 2009 5:31 pm Post subject: Reply with quote
fundamentallysound wrote:
BobboFitos wrote:

Opponent OReb% (on a team level, which can be broken down from the 5man to the individual) is probably the measure that should be looked at more critically.


Are you suggesting using Opponent Team OReb% and then running the adjusted plus minus methodology (like Ryan has been doing with shot selection, efg%, etc.) on that stat to determine on an individual level what the player's impact on Opponent Team OReb% is? That would be something that I think would be very valuable and I'd be interested in seeing. You could, of course, also do the converse and run the same methodology for differences in Opponent Team DReb% to see who are the best at suppressing the other teams ability to grab defensive rebounds (which should mean more offensive boards for their team).


Yes, and what you suggest (differences in team DReb%) would be useful as well.
_________________
-Rob
Back to top
View user's profile Send private message AIM Address
tarak



Joined: 20 Aug 2008
Posts: 9


PostPosted: Tue Mar 17, 2009 5:54 pm Post subject: Reply with quote
Very cool, Ryan. Shot location seems like a great piece of context for rebounding numbers, at least to begin with. Lead/Deficit, Time remaining, and shot clock seem like good ideas, too. I don't know if it would be feasible or turn out to be useful at all, but the shooter's FG% might play into the decision to go after a rebound, and hence be a relevant factor when looking at Reb%.

An additional piece might be some estimate of the player's (expected) rebounding role that is (hopefully) more telling than just position -- say, ranking the 10 players on the floor in terms of their height, tallest to shortest (or, um, just ranking the lineups of each team 1-5). I'm wondering, for instance, if Jason Kidd doesn't stand behind the 3-pt line during free throws as often as most point guards do -- ideally the most meaningful way to look at rebounds off of free throw misses would be to measure by position (1st block, 2nd block, etc.), but that data isn't available . . .. This might all be moot once you include things like the affect of teammates and opponenets, since presumably a coach is trying to put out functional and balanced lineups.

For offensive rebounding, the opponent's average pace might also add context (in the case of team's giving up potential offensive rebounds because they know they need to get back on D).

You mention the possibility of looking at (continuous) shooting distance instead of discrete shooting locations, and it made me wonder if the angle of the shot attempt would be a big part of the relationship to rebounding, and whether that would make things too complicated or messy . . ..

Mountain's points also make sense -- it always seems like Russel Westbrook's value is even higher on the Thunder than it would be on a team that wasn't such a poor shooting team, given his ability to offensive rebound.

bastillon: regarding the correlation between player and team rebound rates - http://www.countthebasket.com/blog/2008 ... g-returns/

bobbyfitos and fundamentally sound: on an individual level, I see why stopping an opponent from getting an offensive rebound is more important than getting the defensive rebound, but at the team level they are the same thing, aren't they? Measuring a player's impact on opponent Team OReb% should be the same as measuring a player's impact on his own team's DReb%.
Back to top
View user's profile Send private message Visit poster's website
fundamentallysound



Joined: 18 Jul 2008
Posts: 14
Location: VA

PostPosted: Tue Mar 17, 2009 6:30 pm Post subject: Reply with quote
tarak wrote:

bobbyfitos and fundamentally sound: on an individual level, I see why stopping an opponent from getting an offensive rebound is more important than getting the defensive rebound, but at the team level they are the same thing, aren't they? Measuring a player's impact on opponent Team OReb% should be the same as measuring a player's impact on his own team's DReb%.


In theory, it should be the same thing, but there are instances where a guy bats the ball out of bounds to prevent the other team from getting it, but doesn't actually secure the rebound. These plays are not as useful as defensive rebounds, because possession is not secured and the opposing team retains it, but they are still useful because the likelihood of scoring off an offensive rebound is greater than scoring off of an in-bounds play. Another reason to run both sets of data is that it could help to bring greater clarity to the numbers, because if you examine both sides of it and come out with divergent numbers on an individual then you would know that the truth likely lies somewhere in the middle. This could perhaps help to reduce noise.
Back to top
View user's profile Send private message AIM Address
tHe_pEsTiLeNcE



Joined: 09 Aug 2006
Posts: 104
Location: where you aren't

PostPosted: Tue Mar 17, 2009 10:12 pm Post subject: Reply with quote
One stat that I would like to see vis a vis accuracy of measuring rebounds is one that somehow incorporates somebody's teammates' rebounding abilities. Troy Murphy is a good example of how team mates can make somebody's rebound rate fluctuate. In 2007-2008, Jermaine O'neal, Jeff Foster, Mike Dunleavy, and Ike Diogu, all significant rebounders, played 8751 minutes for Indiana, and Murphy grabbed only 14% of all available rebounds. The following season, only two of those players are on the team, and they're on a pace to play under 2400 minutes. O'neal's replacement, Nesterovic, is not as good of a rebounder, and doesn't play many minutes to boot. Unsurprisingly, Murphy is grabbing 19.4% of rebounds this year. A similar story happened earlier in his career, when he was a top-10 rebounder in the league with golden state, grabbing about 17% of total boards. In 06-07, his rebound rate suddenly dropped off, likely due to the fact that rebounder extraordinaire Andris Biedrins had his minutes doubled. Why did Murphy's rebounds shoot up so much in his sophomore year? It could be because Danny Fortson, a superb rebounder (at the time), was no longer in the rotation. A player's team mates have a huge impact on what their rebound rate is, and, as such, rebound rate can't be thought of as a pure and absolute stat.
_________________
I'm so sick I be terminally ill

tsherkin



Joined: 31 Jan 2005
Posts: 144


PostPosted: Thu Mar 19, 2009 5:00 pm Post subject: Projected Shot Attempts Reply with quote
I don't know if that's really an apt title, it seems a little vague.

I was wondering if anyone had done any work on projecting how many shots a player should be taking and what kind of methodology was involved.

I mean, we can all get a sense of what a player looks like in terms of his TS%, ORTG and whatever but is there any way to project that to a pace-adjusted shots-per-minute (pr per-36) rate?

I was looking at some hot spot data the other day and looking at the frequency with which certain players got to the rim and how well they shot from mid-range, comparing that against their FTA/FGA and all that and the thought came up.

Anyone?
Back to top
View user's profile Send private message
Ryan J. Parker



Joined: 23 Mar 2007
Posts: 396
Location: Charleston, SC

PostPosted: Thu Mar 19, 2009 5:11 pm Post subject: Reply with quote
I haven't seen any work done on this, but this sorta thing interests me.

The type of questions I ask are: can we put a probability distribution on the types of shots (and % of lineup shots) that a player will take when facing a given lineup?

There are other variants related to that, but I think that is sort of what you're looking for.
_________________
I am a basketball geek.
Back to top
View user's profile Send private message Visit poster's website
tsherkin



Joined: 31 Jan 2005
Posts: 144


PostPosted: Thu Mar 19, 2009 5:25 pm Post subject: Reply with quote
Ryan J. Parker wrote:

The type of questions I ask are: can we put a probability distribution on the types of shots (and % of lineup shots) that a player will take when facing a given lineup?


Well, one imagines that this kind of analysis can't be purely statistical, you have to have some idea of the player, I'd guess, and make effective use of the statistical data in the hot spots.

Quote:
There are other variants related to that, but I think that is sort of what you're looking for.


I guess the root of the issue is that I want to be able to look at a player who's playing well off the bench, or even as a starter, and see where he's at in terms of his optimum shot volume. A guy like Lebron, we know he can handle 21 shots a game, for example, but what about the best way to use Anthony Parker or Matt Barnes? Thad Young?

Are these guys getting as many, fewer or more shots than they should?
Back to top
View user's profile Send private message
Ryan J. Parker



Joined: 23 Mar 2007
Posts: 396
Location: Charleston, SC

PostPosted: Thu Mar 19, 2009 5:37 pm Post subject: Reply with quote
Yeah that's something that I've toyed around with, but I haven't come up with anything concrete yet. On the small sample of players I looked at, an increase of FGA / Team FGA doesn't have a relationship to eFG% or other shooting stats. Except for TS% if I recall correctly, but that was a sort of dirty analysis.
_________________
I am a basketball geek.
Back to top
View user's profile Send private message Visit poster's website
tsherkin



Joined: 31 Jan 2005
Posts: 144


PostPosted: Thu Mar 19, 2009 5:41 pm Post subject: Reply with quote
Ryan J. Parker wrote:
Yeah that's something that I've toyed around with, but I haven't come up with anything concrete yet. On the small sample of players I looked at, an increase of FGA / Team FGA doesn't have a relationship to eFG% or other shooting stats. Except for TS% if I recall correctly, but that was a sort of dirty analysis.


Hmm.

This is a little be more qualitative, but what about analyzing the distribution of a player's shots and producing some kind of adjusted linear measure rating his scoring versatility?

Like you know how 82games.com produces for you that semi-vague shot distribution and what-not?

Well, if we break it down into "finishing at the rim," "the low post" "the top of the circle" "jumpers from 15-22 feet", "threes" and then add something to to account for free throws drawn and FT%, then we might get a rating that could give a better idea of how a player scores and if any one area really stands out.
Back to top
View user's profile Send private message
Ryan J. Parker



Joined: 23 Mar 2007
Posts: 396
Location: Charleston, SC

PostPosted: Thu Mar 19, 2009 5:56 pm Post subject: Reply with quote
That's an interesting idea, although I don't think you'd want to try and take that information and put it into a single number. The ultimate goal is to see maybe how a specific player would work with another player. If you've got a bunch of guys that finish around the rim then maybe you want to try and find other players to exploit the 3.
_________________
I am a basketball geek.
Back to top
View user's profile Send private message Visit poster's website
tsherkin



Joined: 31 Jan 2005
Posts: 144


PostPosted: Thu Mar 19, 2009 6:53 pm Post subject: Reply with quote
Ryan J. Parker wrote:
That's an interesting idea, although I don't think you'd want to try and take that information and put it into a single number. The ultimate goal is to see maybe how a specific player would work with another player. If you've got a bunch of guys that finish around the rim then maybe you want to try and find other players to exploit the 3.


Well, that's not something you need to put into a number; it's pretty clear that if you have slashers and post guys that you need distance shooters as well.

And it doesn't have to be one number, it can be a small collection of numbers, just pertaining to one player, so you can get an idea of where most of his scoring opportunities come from and how good he is in each spot.
Back to top
View user's profile Send private message
Ryan J. Parker



Joined: 23 Mar 2007
Posts: 396
Location: Charleston, SC

PostPosted: Thu Mar 19, 2009 6:59 pm Post subject: Reply with quote
Yeah that's good stuff. I've kinda of focused on figuring out how players increase FG% by their presence. What you're talking about is probably the best place to start first.

Sort of like what I did for rebounding but with shooting?
_________________
I am a basketball geek.
Back to top
View user's profile Send private message Visit poster's website
tsherkin



Joined: 31 Jan 2005
Posts: 144


PostPosted: Thu Mar 19, 2009 7:27 pm Post subject: Reply with quote
Yeah, I imagine something like that might be helpful. We've got the FGAs and FGMs for each spot on the court, broken down even further than the low paint/midrange/3pt/ft thing you used for the rebounds, too, so we could get fairly specific.
Back to top
View user's profile Send private message
tarheeljks



Joined: 24 May 2008
Posts: 2


PostPosted: Fri Mar 20, 2009 4:57 am Post subject: Reply with quote
tsherkin wrote:
Ryan J. Parker wrote:
Yeah that's something that I've toyed around with, but I haven't come up with anything concrete yet. On the small sample of players I looked at, an increase of FGA / Team FGA doesn't have a relationship to eFG% or other shooting stats. Except for TS% if I recall correctly, but that was a sort of dirty analysis.


Hmm.

This is a little be more qualitative, but what about analyzing the distribution of a player's shots and producing some kind of adjusted linear measure rating his scoring versatility?

Like you know how 82games.com produces for you that semi-vague shot distribution and what-not?

Well, if we break it down into "finishing at the rim," "the low post" "the top of the circle" "jumpers from 15-22 feet", "threes" and then add something to to account for free throws drawn and FT%, then we might get a rating that could give a better idea of how a player scores and if any one area really stands out.



so you essentially are looking to create a more detailed version of the "shooting details" on 82games.com?
Back to top
View user's profile Send private message AIM Address
tsherkin



Joined: 31 Jan 2005
Posts: 144


PostPosted: Fri Mar 20, 2009 7:02 am Post subject: Reply with quote
With some element of a rating system associated with the regional details.

Eli W



Joined: 01 Feb 2005
Posts: 402


PostPosted: Wed Mar 18, 2009 10:14 am Post subject: When Losing Leads to Winning Reply with quote
There's some pretty interesting discussion going on springing from a research paper that suggests that it could be a good thing to be trailing by one point at the half of a college basketball game.

Here's the NY Times piece:

http://www.nytimes.com/2009/03/16/sport ... .html?_r=2

It points to the original paper:

http://qbox.wharton.upenn.edu/documents ... inning.pdf

Some criticism of the study can be found on various blogs (the second link in particular may be of interest as it follows up with some NBA research):

http://www.insidethebook.com/ee/index.p ... ood_thing/

http://www.insidethebook.com/ee/index.p ... g_part_ii/

http://www.stat.columbia.edu/~cook/mova ... _news.html

http://freakonomics.blogs.nytimes.com/2 ... o-winning/
Back to top
View user's profile Send private message Send e-mail Visit poster's website
Ryan J. Parker



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

PostPosted: Wed Mar 18, 2009 10:22 am Post subject: Reply with quote
Gelman's analysis caught my eye this morning. I love this stuff. Right or wrong, this is a great learning tool when really smart people analyze this kinda thing.
_________________
I am a basketball geek.
Back to top
View user's profile Send private message Visit poster's website
Guy



Joined: 02 May 2007
Posts: 128


PostPosted: Wed Mar 18, 2009 10:40 am Post subject: Reply with quote
Does anyone here know of NBA data that would show if the same pattern occurs there? Ed K. posted some data on win% at various minute/score states a while back, but the link is now broken.
Back to top
View user's profile Send private message
Ed Küpfer



Joined: 30 Dec 2004
Posts: 786
Location: Toronto

PostPosted: Wed Mar 18, 2009 2:17 pm Post subject: Reply with quote
I wish I could help. The original data is gone, disappeared on a long lost hard drive. I now have newer, better data, which I am obviously not allowed to give out.

I can tell you this: my original dataset came from NBA.com play by plays. You can recreate this by using the data provided at basketballvalue. In fact, it will be much better since their data is much cleaner than the stuff I used. I don't know how hard it would be to do this, but if I can do it, anyone can do it in half the time.
_________________
ed
Back to top
View user's profile Send private message Send e-mail
Guy



Joined: 02 May 2007
Posts: 128


PostPosted: Wed Mar 18, 2009 3:34 pm Post subject: Reply with quote
Ed: Can you just tell us how often teams win when they lead by 1 point at the half? Surely that isn't a proprietary figure. :>)
Back to top
View user's profile Send private message
Ed Küpfer



Joined: 30 Dec 2004
Posts: 786
Location: Toronto

PostPosted: Wed Mar 18, 2009 4:01 pm Post subject: Reply with quote
I found some old data that may be helpful. Home team halftime leads and eventual wins, games from 1992 to 2002. (The 20 point leads and deficits, I'm not sure if they mean actual 20 points or 20+ points.)

Code:
lead wins games
-20 0 43
-19 4 55
-18 5 67
-17 6 72
-16 4 81
-15 14 117
-14 18 118
-13 32 155
-12 39 164
-11 38 221
-10 63 236
-9 64 246
-8 112 311
-7 113 364
-6 120 324
-5 147 386
-4 175 401
-3 190 395
-2 241 479
-1 255 460
0 271 497
1 296 509
2 289 483
3 322 500
4 335 513
5 315 438
6 372 462
7 331 424
8 311 377
9 293 358
10 296 355
11 293 336
12 266 287
13 245 279
14 205 231
15 207 221
16 169 184
17 178 185
18 135 140
19 115 116
20 102 105

_________________
ed
Back to top
View user's profile Send private message Send e-mail
Serhat Ugur (hoopseng)



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

PostPosted: Wed Mar 18, 2009 4:11 pm Post subject: Reply with quote
last season it is 52-56. (trailing by 1 and win the game)
this season it is 38-47
_________________
http://www.nbastuffer.com

Last edited by Serhat Ugur (hoopseng) on Wed Mar 18, 2009 4:27 pm; edited 1 time in total
Back to top
View user's profile Send private message Visit poster's website
Guy



Joined: 02 May 2007
Posts: 128


PostPosted: Wed Mar 18, 2009 4:24 pm Post subject: Reply with quote
Thanks, Ed. Very helpful.
Back to top
View user's profile Send private message
Mike G



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

PostPosted: Thu Mar 19, 2009 6:11 am Post subject: Reply with quote
I'm guessing Ed's data must be referring to the halftime status of the home team. A halftime tie is won 56% of the time by somebody, eh?. Likewise:
Code:
lead win
+20 .982
+15 .914
+10 .841
+5 .723
0 .561
-5 .398
-10 .235
-15 .114
-20 .041

In each case, this team does much better in the 2nd half; and a halftime deficit is much worse for the 'other' team.
_________________
`
36% of all statistics are wrong
Back to top
View user's profile Send private message Send e-mail
Guy



Joined: 02 May 2007
Posts: 128


PostPosted: Thu Mar 19, 2009 9:45 am Post subject: Reply with quote
Hoopseng: Is that data for all teams, or home teams only?
Back to top
View user's profile Send private message
schtevie



Joined: 18 Apr 2005
Posts: 412


PostPosted: Thu Mar 19, 2009 10:41 am Post subject: Reply with quote
The paper is cute, but I don't find it that interesting from a basketball perspective. And apparently not at all from an NBA perspective.

Given Ed's data, and the familiar home court advantage of between 3 to 4 points - call it half that at halftime, it would seem that, lo and behold, home teams down 2 points at the half win....half their games (241 out of 479). Maybe this has something to do the attitudes of professionals vs. amateurs. Lack of rah, rah, and all that.

More generally, It would have been interesting if the study had included corroborating analysis of overtime games. According to their set-up, as I understand it, a game that ends in a tie is actually a very slight notional deficit for the away team, given the generally accepted existence of the home court advantage, and it should invoke their relatively greater effort. It would have been nice confirmation to see away teams winning more than 50% of OT games. Maybe it is so, maybe it isn't.
Back to top
View user's profile Send private message
Guy



Joined: 02 May 2007
Posts: 128


PostPosted: Thu Mar 19, 2009 11:21 am Post subject: Reply with quote
Actually, the NBA data provides partial support. In Ed's data, and more recent data compiled by Mitchell Lichtman, it appears that HOME teams down one at the half win more than they should. They win about as frequently (slightly more) than in tie games. (Alternatively, road teams up by one at the half win less than they should.) Road teams do not show this -1 effect. See the 2nd link in Eli's original post above for discussion of this at The Book blog.
Back to top
View user's profile Send private message
Ed Küpfer



Joined: 30 Dec 2004
Posts: 786
Location: Toronto

PostPosted: Thu Mar 19, 2009 12:27 pm Post subject: Reply with quote
Okay, I went ahead and did some more work on this. By incorporating the home and away teams' win percentage, I calcuated the home team win probabilty (log5) for each game in the old data I have. I created a histogram of home team half time leads, binned by the probability of the home team winning the game.



For some reason, "home team down one at the half" games are overrepresented when the home team is playing a far superior opponent.

Next home team win percentages by lead at the half, again binned by log5:



Nothing looks out of line there. It appears that once you take relative team strengths into account, home teams down one at the half win about as much as they're supposed to.
_________________
ed
Back to top
View user's profile Send private message Send e-mail
ecumenopolis0



Joined: 15 Jul 2008
Posts: 22
Location: Houston

PostPosted: Thu Mar 19, 2009 12:33 pm Post subject: Reply with quote
schtevie wrote:
It would have been nice confirmation to see away teams winning more than 50% of OT games. Maybe it is so, maybe it isn't.


Since the 86-87 season, home wins OT games 796-741, or at about .518

Home/away may not be the best OT predictor.

Somewhat OT (off-topic, not overtime), but it was always my understanding that the team to force overtime with that last score or stop in regulation was much more likely to win. I have no numbers to back this up, but this would seem to me the best way to predict the winner.
Back to top
View user's profile Send private message AIM Address
Ed Küpfer



Joined: 30 Dec 2004
Posts: 786
Location: Toronto

PostPosted: Thu Mar 19, 2009 12:55 pm Post subject: Reply with quote
Further evidence that I shouldn't trust my eyes to see patterns in the data. I regressed probability of home team win on home and away team win percentages along with home team lead at the half, the latter variable treated as a categorical (factor in R parlance). Here are the regression results, which do show something fishy about -1, but I'm not sure what.

Code:
coef.est coef.se
a.win.p -0.68 0.02
h.win.p 0.76 0.03
factor(h2)-15 0.17 0.04
factor(h2)-14 0.23 0.04
factor(h2)-13 0.25 0.04
factor(h2)-12 0.28 0.04
factor(h2)-11 0.21 0.03
factor(h2)-10 0.30 0.03
factor(h2)-9 0.31 0.03
factor(h2)-8 0.38 0.03
factor(h2)-7 0.32 0.03
factor(h2)-6 0.38 0.03
factor(h2)-5 0.39 0.03
factor(h2)-4 0.42 0.03
factor(h2)-3 0.46 0.03
factor(h2)-2 0.49 0.03
factor(h2)-1 0.54 0.03 <-- Oy!
factor(h2)0 0.52 0.03
factor(h2)1 0.54 0.03
factor(h2)2 0.56 0.03
factor(h2)3 0.58 0.03
factor(h2)4 0.59 0.03
factor(h2)5 0.65 0.03
factor(h2)6 0.73 0.03
factor(h2)7 0.71 0.03
factor(h2)8 0.74 0.03
factor(h2)9 0.73 0.03
factor(h2)10 0.74 0.03
factor(h2)11 0.77 0.03
factor(h2)12 0.79 0.03
factor(h2)13 0.77 0.03
factor(h2)14 0.78 0.03
factor(h2)15 0.83 0.03
---
n = 10647, k = 33
residual deviance = 1786.7, null deviance = 6268.0 (difference = 4481.3)
overdispersion parameter = 0.2
residual sd is sqrt(overdispersion) = 0.41


(I only included games where the absolute lead was less than 16.)
_________________
ed

hoopseng



Joined: 13 Oct 2006
Posts: 115
Location: Basketball Research

PostPosted: Fri Mar 20, 2009 12:38 am Post subject: Reply with quote
Guy wrote:
Hoopseng: Is that data for all teams, or home teams only?

All teams
_________________
http://www.nbastuffer.com
Back to top
View user's profile Send private message Visit poster's website
hoopseng



Joined: 13 Oct 2006
Posts: 115
Location: Basketball Research

PostPosted: Fri Mar 20, 2009 12:45 am Post subject: Reply with quote
[quote="ecumenopolis0"]
schtevie wrote:


Somewhat OT (off-topic, not overtime), but it was always my understanding that the team to force overtime with that last score or stop in regulation was much more likely to win. I have no numbers to back this up, but this would seem to me the best way to predict the winner.


I didn't search the board, but I could remember, in a discussion that we have been at here: the first team to score in OT, more likely to win the game.
_________________
http://www.nbastuffer.com
Back to top
View user's profile Send private message Visit poster's website
Guy



Joined: 02 May 2007
Posts: 60


PostPosted: Fri Mar 20, 2009 9:24 am Post subject: Reply with quote
Interesting data from Ed. I can think of two possible reasons for the -1 anomaly. First, and less important, is the distribution of scores in the second half. In games where talent disparity is small (many of these games), the most common score is probably home +1. Given a normal distribution around +1 or +1.5, and assuming home team wins about 54% in OT, that would give you a pretty big jump from in win% from -2 to -1.

Second, and likely more important, is that the superior team in basketball has a true ability to outscore its opponent greater than the actual game outcomes. Where would we expect this to manifest itself the most. Well, in close games more than blowouts, of course. And I would expect it to peak at around -1, where the better team expressing its full superiority has the greatest payoff.

Ed, can you rerun your last set of #s showing not the home team win% but rather the superior team? I think we may see the same discontinuity for strong teams.
Back to top
View user's profile Send private message
Ryan J. Parker



Joined: 23 Mar 2007
Posts: 396
Location: Charleston, SC

PostPosted: Fri Mar 20, 2009 10:12 am Post subject: Reply with quote
More good discussion:

http://www.stat.columbia.edu/~cook/mova ... e_que.html

http://freakonomics.blogs.nytimes.com/2 ... -response/