APBRmetrics

Mike G wrote:If you play 23 postseason games and lose at the Finals, is that just as good as playing 23 games, winning it all, and having nobody else to play?
At one time there were just 2 rounds -- the Russell Celtics would beat one opponent to get into the Finals.
If there had been more rounds, they'd have gotten into more playoff games. Since the league has expanded, later players can get into a lot more games.

If that's your goal your bonus should apply only to the early NBA years, no?
Because if you apply it to Russell Celtics and current teams how exactly does it help players from the 60s?

Mike G wrote:The bonus of 10 equivalent playoff games assumes that another round would go on avg 5.5 games; a champion would have a 50/50 chance of yet another 5.5 games; and so on. These possible series add up to 11. The number 10 just rounds it down a bit, since we never really have an infinite tournament.

But if you want to go down this road you shouldn't then adjust players' career values for being on bad teams [for missing playoffs games] and really good teams [for playing in the playoffs more]? Hell, you can easily make an argument that almost all top players in history could have played more if not for injuries, conditions, slow league evolution etc which is why I don't understand this bonus, you made up a rule which has nothing to do with their actual statistical achievements.
Don't get me wrong, it's your list and you can make whatever rules you want about it, I just find it inconsistent.

Mike G wrote:Robert Horry got lots of playoff minutes and titles. Clifford Robinson did not. Their po/rs ratios are at the opposite ends of the individual success spectrum, and this definitely affected the fortunes of their teams.

But this difference is already captured in their production in the playoffs. So by counting titles too you include this information twice.

Mike G wrote:Awards, meanwhile, supposedly go to the best players. If you're among the best at something, it's usually contained in your stats. A major motivation to make a list that's 100% statistically determined is that I didn't want to perpetuate the hype -- or conversely, continue to overlook a player's excellence. Kobe's rookie all-star game, and his selection this past season, do nothing to elevate his stature.

I agree with this approach but I find it weird that you decided to add one little bonus based on a team award.

Mike G wrote:

... how many good years would LeBron need to overtake Jordan as #1?

- 28 games to catch Karl Malone
- 73 games to catch Shaq at #5

So he will be in the Top5 all-time at the age of 30...

Mike G wrote:These numbers assume he hasn't won another title in the meantime. To pass Jordan to the top spot, he can continue his career rates for :
- 346 games with no title.
- 330 games with a 3rd title
- 314 G with 2 more titles.

... with possible GOAT title at the age of 33.

I recently tried to rank players' careers myself [with slightly different approach and goal] and my main takeaway was that current players have huge advantage because of longer career's length and that we somehow underestimate historical greatness of LeBron because he still plays so his totals don't jump at us as they should based on his age.
I mean seriously, it's insane that next season he can catch Malone at #6 who played 600+ more games which is above average career in itself!

wilq wrote:

Mike G wrote:If you play 23 postseason games and lose at the Finals, is that just as good as playing 23 games, winning it all, and having nobody else to play?
At one time there were just 2 rounds -- the Russell Celtics would beat one opponent to get into the Finals.
If there had been more rounds, they'd have gotten into more playoff games. Since the league has expanded, later players can get into a lot more games.

If that's your goal your bonus should apply only to the early NBA years, no?
Because if you apply it to Russell Celtics and current teams how exactly does it help players from the 60s?

In the same year: The 2001 Lakers went 15-1; the Sixers went 12-11. Is 23 games better than 16?
The Lakers were the much better team -- basically, their players were better. The ratio 26/23 seems to fit better than 16/23.

From one era to another: The '61 and '64 Celtics won the title with 10 postseason games; the 2008 Celts went 26 games for theirs. The disparity of 26/10 seems to be greater than 36/20.

The '60s players are 'helped' by the fact that their chance of winning a title in a given year were 1/8 or 1/9, rather than 1/27 or 1/30. That's not how it worked out, since the Celts won all but one; but outside of 1957-1969 it's relevant. And from '68 thru '76, with ABA, two teams won titles.

... I find it weird that you decided to add one little bonus based on a team award....

A championship isn't an award; it's an accomplishment. The fact that every other team, after winning a series, gets to play more games, is statistically a punishment for the champions. The 'bonus' is an adjustment to correct that inconsistency.

Mike G wrote: Robert Horry got lots of playoff minutes and titles. Clifford Robinson did not. Their po/rs ratios are at the opposite ends of the individual success spectrum, and this definitely affected the fortunes of their teams.

But this difference is already captured in their production in the playoffs. So by counting titles too you include this information twice.

You're somewhat correct. But a player who is explosive and inconsistent, somehow failing to win the close ones, can have better numbers than the guy who has a knack for coming up with a big game when needed -- or hitting the buzzer beater. So 'winning' is just slightly quantifiable.

Robert Horry ranks 150th on my list. He's among Spencer Haywood, Mel Daniels, Jason Terry, Johnny Kerr.
With no extra credit for his 7 rings, he drops to #173 -- among Mark Price, Billy Paultz, Kevin Willis, Rashard Lewis -- and 3 spots below Cliff Robinson.
With no extra weight on playoff games, he's all the way down at #338; in the company of Mike Miller, Orlando Woolridge, Keith Van Horn. About what you'd expect from someone who was never an allstar, who started just 43% of his games, and who averaged 25 minutes.

There are other adjustments that I haven't mentioned:
- Old timers who played short schedules are given their 82-game equivalents: In a 72-game season, their 'equivalent totals' are factored by 82/72.
- Until the early '70s, league Ast/FG ratio was considerably lower than it is now. I've boosted the Ast rates of players, season by season, before the modern Ast/FG ratio came along. Basically, I split the difference.
- ABA/NBA equivalents are factored, also year by year, for Sco, Reb, Ast rates. In the first years of the ABA, points and rebounds were pretty easy to get.

Thanks for the grilling. Most likely, others have wondered some of the same things.

wilq wrote:...
I recently tried to rank players' careers myself [with slightly different approach and goal] and my main takeaway was that current players have huge advantage because of longer career's length and that we somehow underestimate historical greatness of LeBron because he still plays so his totals don't jump at us as they should based on his age.
I mean seriously, it's insane that next season he can catch Malone at #6 who played 600+ more games which is above average career in itself!

Longer careers are possible. But some would counter that the competition is greater. Specifically, the need to be a super-athlete is paramount. The average player is in better shape today, but he's asked to do more physically demanding stuff. So, while he may not have a career- ending injury, a loss in quickness can cost him his job.

Kobe's athleticism worked for him a good long while. Then he moved so quickly he just broke.
Dominique Wilkins' muscles also tore his own leg apart, in an explosive step.

LeBron is a physical marvel, and yet no one is indestructible. Karl Malone broke suddenly and irreversibly after an iron-man career.
Every super-season after age 30 or so is something of a miracle.

While the Russell Celtics get the lion's share of those 10-game championship bonuses in the early years, most players' rankings are not seriously affected by them. Largely, it's just those Celtics.
On the left, players who benefit most -- by woB/wB -- from their statistical Ring bonuses.
On the right, players with no (or few) rings who are dropped below others with (more) rings.
Columns are for championships (NBA and ABA), rank with, and without, title bonus.Wilt and Kareem are very close, and 6 > 2 tips the scale.
Garnett's 1 title is overtaken by Magic's 5.
Robinson got 2 rings, but Russell had 11.

Mike G wrote:In the same year: The 2001 Lakers went 15-1; the Sixers went 12-11. Is 23 games better than 16? The Lakers were the much better team -- basically, their players were better. The ratio 26/23 seems to fit better than 16/23.

If your focus was on a single season, I would buy that argument.
But you try to measure entire careers and you already include information that Lakers were better than Sixers in the early 2000s. How? Lakers played less playoff games in 2001 but way more such games in later seasons! So Kobe/Shaq accumulated more stats in the playoffs than Iverson despite such unusual scenario in 2001.

Mike G wrote:From one era to another: The '61 and '64 Celtics won the title with 10 postseason games; the 2008 Celts went 26 games for theirs. The disparity of 26/10 seems to be greater than 36/20. [...] The '60s players are 'helped' by the fact that their chance of winning a title in a given year were 1/8 or 1/9, rather than 1/27 or 1/30.

Isn't it another argument against your method? Celtics in the 60s played fewer playoff games so their career numbers were lower but their road to the championship was much, much easier. So why Russell's bonus is equal to Garnett's bonus?

Mike G wrote:A championship isn't an award; it's an accomplishment. The fact that every other team, after winning a series, gets to play more games, is statistically a punishment for the champions.

I disagree because winning side should have better stats so the difference in outcome is already included in the players' numbers.

BTW, Let me rephrase my main objection...

Mike G wrote:No awards are considered in this summary -- just on-court stats

Is championship ring an on-court statistic?

Mike G wrote:But a player who is explosive and inconsistent, somehow failing to win the close ones, can have better numbers than the guy who has a knack for coming up with a big game when needed -- or hitting the buzzer beater.

What's wrong with that? I thought your goal was to create pure statistical rankings in which better numbers should mean that player had a better career. Now it seems like you wanted a correction because you weren't happy with the results so you added an adjustment based on rings... which is a deviation from your stated goal of those rankings.

So Kobe/Shaq accumulated more stats in the playoffs than Iverson despite such unusual scenario in 2001.

Did they?
http://www.basketball-reference.com/pla ... _2001.html
Iverson was the 2001 postseason leader in Min., FG, FGA, 3fga, FT, FTA, Ast, Stl, TO, and Pts
Mutombo led in OReb, DReb, and Blk
Tyrone Hill led in PF, and that is every stat total category.

Win Shares were: 3.8 Kobe, 3.8 Mutombo, 3.7 Shaq

The statistics accumulated over 10 to 25 games by a champion do not give any benefit to the winner of the final series; just to every series winner up to that point.
It's possible for the Finals winner to be outscored by the loser, so it's not a given that they have more of anything, even for that series.
I'd guess that in general, the champions have played fewer games than the runners-up.

The "Ring" is a (short) column header representing an NBA title, a championship, a playoff series after which there are no more series. The 10-game stat-equivalent bonus seems like a perfectly reasonable statistical bonus. Sometimes it's less perfect than other times; always it's better than nothing.

It's pretty much analogous to the extrapolation of 72-game totals to 82-game equivalents.
Or the quantitative recognition that assists were once less generously granted. Or that the ABA was less competitive in the beginning than it was at the end.

In recent seasons, I've adjusted players' assists downward (mostly) based on their home scorekeepers' generosity relative to the rest of the league.
I also adjust them upward, based on the % of their teammates' FG which are 3's.
At no time do I consider "raw stats" to be more pure than after adjusting them to historic standards. They're skewed when they hit the boxscore. The whole idea is to somewhat level the playing field for different scoring/rebounding/assisting milieus.

Celtics in the 60s played fewer playoff games so their career numbers were lower but their road to the championship was much, much easier. So why Russell's bonus is equal to Garnett's bonus?

That's a very valid question, and I don't have a pat answer. Subjectively, I like for championships to be recognized. I could dilute the NBA and ABA champs' credits. But I don't mind if the old-timers get credit for some non-quantifiable stuff, like:
- playing for lousy money under lousy conditions, both physical and emotional.
- overcoming racism, segregation, exploitative schools and agents and rules
- the drug scourge, which accounts for a huge number of very short careers in the '70s-'80s

Garnett gets to come to the league at age 18, and he'll play 18-20 years.
A guy coming into a monopoly league (of 8-10 teams) at age 22 may play 8 to 10 years.
Both guys have roughly equal chance to win a ring.

When I sort the rankings by decade, or by year of entry, I get numbers of players -- in the top 50, top 150, etc -- roughly proportional to the number of teams in the league. So it doesn't seem as if the rankings favor one era over another -- assuming that league expansion has just about kept up with the talent pool.

I've made a faux eWins creating the closest career regular-season match between players in my top 100 and Win Shares (b-r.com)
The formula is : eW = Min*(T-13.3)/6000
T is the standardized/weighted total of other rates, as shown in the OP list.
This is just for the purpose of the current post, which is to make playoff eW weigh more heavily than RS eWins, and see how such lists compare to the OP.
Just adding RS+PO eW, we see that such a ranking favors a long and productive career (1st set of columns)
Adding a multiple of PO eW to RS eW, the ranking changes to favor great postseasons.I'm calling these 'credits', since they don't represent actual eWins. After adding the PO*N, I've normalized the whole list of 708 players to the same constant value of RS+PO.
Oscar and Moses are the only top 20 players (in the PO*1 column) whose PO/RS eW are less than average (.075); so they quickly drop off the list.
Others with less than great postseasons get more credits with higher multiples, but still lose ground to others.
I think a multiplier of 10 is rather unconscionable; but Kareem outranks Jordan with lower multiples.

Mike G wrote:

So Kobe/Shaq accumulated more stats in the playoffs than Iverson despite such unusual scenario in 2001.

Did they? [...] Win Shares were: 3.8 Kobe, 3.8 Mutombo, 3.7 Shaq

To clarify, I didn't mean "more stats in the *2001* playoffs" but total for their careers.

Mike G wrote:The statistics accumulated over 10 to 25 games by a champion do not give any benefit to the winner of the final series;

IMO it creates the advantage over all players who lost earlier because they had more chances to accumulate stats.

Mike G wrote:I'd guess that in general, the champions have played fewer games than the runners-up.

But they have higher chance for more playoffs games in the future, don't they?

Mike G wrote:Garnett gets to come to the league at age 18, and he'll play 18-20 years.
A guy coming into a monopoly league (of 8-10 teams) at age 22 may play 8 to 10 years.
Both guys have roughly equal chance to win a ring.

How did you calculate it? Because I strongly disagree with the conclusion.
If my goal is rings, I'd rather play 8-10 years close to my peak against lower number of opponents than what KG had to go through.

Mike G wrote:When I sort the rankings by decade, or by year of entry, I get numbers of players -- in the top 50, top 150, etc -- roughly proportional to the number of teams in the league. So it doesn't seem as if the rankings favor one era over another -- assuming that league expansion has just about kept up with the talent pool.

Wait a second, in your opinion such ranking should equally represent players from all the decades?

No, the number of players from any year (or decade), in the to 100, top 200, etc, is roughly proportional to the number of NBA teams at that time.

If my goal is rings, I'd rather play 8-10 years close to my peak against lower number of opponents...

With fewer opponents, you will likely get more rings but play fewer playoff games. How to rectify this? In part, by giving a bonus for championships.

Mike G wrote: I'd guess that in general, the champions have played fewer games than the runners-up.

But they have higher chance for more playoffs games in the future, don't they?

I'd guess the correlation is minimal. Is there a study done to this effect?
It's not better to assume a player's reward for this year is sometime in the future. Frankly, the suggestion is baffling.

IMO it creates the advantage over all players who lost earlier because they had more chances to accumulate stats.

That is true up until the Finals. And then, the winners get as many more chances as the loser -- none.