2010-11 eWins leaders

[Mike G]

A bit out of sequence, but revived from the recent meltdown.
Estimated equivalent wins added to an average team.
Rates are per 36 minutes and per 100 points and 44 rebounds per team. Your top 50-something:

Code: Select all

eWins   per36 rates     tm   Min   Eff%   Sco   Reb   Ast    PF   Stl   TO   Blk   e484
17.7   James,LeBron    Mia   39   .581   29.4   7.9   7.6   2.0   1.5   3.5   .6   2.79
15.6   Howard,Dwight   Orl   38   .594   26.7  15.4   1.5   3.3   1.4   3.5  2.3   2.57
15.4   Rose,Derrick    Chi   37   .540   28.6   4.6   8.2   1.7   1.0   3.4   .6   2.45
14.3   Wade,Dwyane     Mia   37   .568   29.0   7.1   5.2   2.6   1.5   3.1  1.1   2.46
14.0   Bryant,Kobe     LAL   34   .538   30.5   6.0   5.5   2.3   1.3   3.3   .2   2.43

13.5   Griffin,Blake   LAC   38   .535   22.3  13.0   3.5   3.0    .8   2.7   .5   2.11
13.4   Durant,Kevin    Okl   39   .576   28.4   7.0   2.7   1.9   1.1   2.7   .9   2.13
13.1  Westbrook,Russel Okl   35   .526   23.9   5.4   9.3   2.7   2.0   4.2   .4   2.22
12.8   Gasol,Pau       LAL   37   .577   21.5  10.8   3.6   2.5    .6   1.7  1.5   2.04
12.1   Paul,Chris      NOH   36   .566   18.7   4.8  10.0   2.5   2.4   2.3   .1   2.04

eWins   per36 rates     tm   Min   Eff%   Sco   Reb   Ast    PF   Stl   TO   Blk   e484
12.1   Love,Kevin      Min   36   .579   21.0  16.6   2.4   2.1    .6   2.2   .4   2.23
12.0  Aldridge,Lamarcu Por   40   .540   22.4   9.3   2.0   2.5   1.0   1.8  1.1   1.81
11.9   Randolph,Zach   Mem   36   .543   22.1  14.0   2.3   2.4    .9   2.1   .3   2.11
11.8   Nowitzki,Dirk   Dal   34   .600   28.9   8.2   3.0   2.6    .6   2.1   .7   2.28
11.4  Stoudemire,Amare NYK   37   .553   25.3   8.4   2.8   3.6    .9   3.3  2.0   1.93

10.7   Pierce,Paul     Bos   35   .607   25.1   6.3   3.6   3.1   1.1   2.3   .7   1.86
10.7   Jefferson,Al    Uta   36   .522   19.4  11.0   1.9   3.0    .6   1.3  1.9   1.75
9.9    Ginobili,Manu   SAS   30   .570   23.7   4.9   6.4   2.5   1.9   2.7   .4   1.98
9.9    Horford,Al      Atl   35   .580   18.4  10.8   3.6   2.7    .8   1.7  1.1   1.77
9.8    Bosh,Chris      Mia   36   .557   21.4   9.4   2.2   2.3    .8   1.8   .6   1.69

eWins   per36 rates     tm   Min   Eff%   Sco   Reb   Ast    PF   Stl   TO   Blk   e484
9.8   Martin,Kevin     Hou   33   .586   28.4   3.9   2.9   2.2   1.2   2.7   .2   1.81
9.6   Parker,Tony      SAS   32   .560   22.0   3.8   8.2   2.0   1.3   3.0   .0   1.84
9.6   Garnett,Kevin    Bos   31   .567   21.2  11.7   3.1   2.5   1.6   2.0   .9   2.09
9.5   Smith,Josh       Atl   34   .531   19.3  10.1   3.5   3.1   1.4   2.8  1.7   1.74
9.4   Ellis,Monta      GSW   40   .528   21.3   3.2   4.8   2.3   1.9   2.9   .3   1.41

9.3   Nash,Steve       Phx   33   .591   17.3   4.0  12.7   1.3    .7   4.0   .1   1.79
9.2   Duncan,Tim       SAS   28   .529   18.8  12.8   3.8   2.1    .9   2.2  2.4   2.07
9.1   West,David       NOH   35   .549   22.6   9.1   2.5   3.1   1.0   2.2   .9   1.79
8.8   Millsap,Paul     Uta   34   .568   20.0   9.1   2.7   3.9   1.5   2.1  1.0   1.63
8.8   Odom,Lamar       LAL   32   .581   18.1  10.1   3.6   2.7    .6   1.9   .8   1.61

eWins   per36 rates     tm   Min   Eff%   Sco   Reb   Ast    PF   Stl   TO   Blk   e484
8.7   Lopez,Brook      NJN   35   .538   22.5   6.9   1.5   3.1    .6   2.3  1.5   1.46
8.5   Deng,Luol        Chi   39   .540   18.9   6.2   2.9   1.9    .9   1.8   .5   1.28
8.4   Scola,Luis       Hou   33   .531   20.8   9.9   3.1   3.6    .8   2.3   .7   1.69
8.4   Granger,Danny    Ind   35   .544   22.6   5.9   2.7   2.9   1.2   2.8   .8   1.47
8.3   Brand,Elton      Phi   35   .545   17.5   9.5   1.6   3.3   1.2   1.4  1.4   1.43

8.2   Boozer,Carlos    Chi   32   .534   23.5  12.8   3.2   3.7    .9   3.0   .3   2.10
8.0   Williams,Deron   Uta   38   .575   22.3   4.2   9.2   2.8   1.2   3.5   .2   1.92
7.9   Hilario,Nene     Den   31   .638   19.9   9.9   2.7   3.9   1.4   2.2  1.1   1.67
7.8   Curry,Stephen    GSW   34   .589   21.3   4.3   6.1   3.5   1.6   3.4   .3   1.52
7.8   Miller,Andre     Por   33   .520   15.7   4.9   8.2   2.4   1.6   2.8   .2   1.42

eWins   per36 rates     tm   Min   Eff%   Sco   Reb   Ast    PF   Stl   TO   Blk   e484
7.8   Rondo,Rajon      Bos   37   .490   11.5   4.8  12.0   1.8   2.2   3.5   .2   1.49
7.7   Allen,Ray        Bos   36   .607   21.0   3.9   3.0   1.9   1.0   1.5   .2   1.30
7.4   Lee,David        GSW   36   .541   16.6  10.1   3.1   3.0   1.1   2.4   .4   1.36
7.4   Anthony,Carmelo  Den   35   .536   26.4   8.5   3.2   2.8    .9   3.0   .6   2.02
7.2   Johnson,Joe      Atl   35   .512   20.1   4.6   4.8   1.9    .7   2.1   .1   1.37

7.2   Holiday,Jrue     Phi   35   .519   15.4   4.5   6.8   2.6   1.5   2.8   .4   1.20
7.1   Conley,Mike      Mem   35   .514   14.9   3.6   6.9   2.4   1.9   2.3   .2   1.19
7.0   Iguodala,Andre   Phi   37   .520   14.9   6.1   6.3   1.7   1.5   2.1   .5   1.38
6.8   Lowry,Kyle       Hou   34   .541   14.7   4.7   7.6   3.0   1.5   2.3   .3   1.29
6.8   Nelson,Jameer    Orl   31   .542   18.1   4.1   8.1   3.4   1.2   3.2   .0   1.42
6.8   Gasol,Marc       Mem   32   .567   15.3   9.2   2.9   3.9   1.1   2.1  1.9   1.27
6.8   Bogut,Andrew     Mil   35   .488   14.3  12.5   2.2   3.5    .8   2.0  2.6   1.43

This edition does not total players who were with multiple teams.
There are 24 players above 1.80 eWins per 484 minutes; so that might be considered All-Star level.
Eff% = Pts/(FGA*2 + FTA)

[Mike G]

Here's a look at the fraction of teams' minutes which were had by players classed according to their eWins/484 (win-producing) rate.
A rate above 2.0 might be called superstar, above 1.5 is a star.
Above 1.0 is above average, and below .50 is 'marginal'.
An e484 rate of zero is 'replacement level'.

Code: Select all

Playoff Teams - fraction of minutes above various eW/484 levels
East
e484    Chi    Mia    Bos    Orl    Atl    NYK    Phl    Ind      Avg
2.0     .25    .30    .11    .15    .00    .05    .00    .00      .11
1.5     .33    .44    .29    .15    .27    .23    .00    .00      .21
1.0     .49    .44    .59    .57    .40    .42    .69    .53      .52
.50     .87    .52    .83    .85    .74    .80    .97    .75      .79
.00     .99    .92    .97    .95    .92    .96    .99   1.00      .96
                                    
West    SAS    Dal    LAL    Okl    Den    NOH    Por    Mem      Avg
2.0     .11    .13    .29    .30    .09    .15    .00    .14      .15
1.5     .36    .23    .50    .30    .21    .27    .20    .14      .28
1.0     .45    .68    .50    .41    .58    .38    .55    .60      .52
.50     .96    .86    .76    .70    .92    .76    .88    .87      .84
.00     .99   1.00    .97    .99    .97    .99    .98    .97      .98
 

Non-playoff teams  
 East                                
e484    Cha    Cle    Det    Mil    NJN    Tor    Was      Avg    
2.0     .00    .00    .00    .00    .00    .00    .00      .00    
1.5     .00    .00    .00    .00    .00    .00    .00      .00    
1.0     .44    .39    .48    .42    .36    .32    .35      .39    
.50     .86    .53    .94    .81    .71    .86    .73      .78    
.00     .96    .97    .99   1.00    .93    .98    .94      .97    
                                     
West    GSW    Hou    LAC    Min    Phx    Sac    Uta      Avg    
2.0     .00    .00    .16    .13    .00    .00    .00      .04    
1.5     .13    .26    .16    .13    .17    .00    .38      .17    
1.0     .42    .51    .41    .27    .32    .27    .51      .39    
.50     .74    .88    .66    .47    .85    .85    .66      .73    
.00     .99   1.00    .96    .99   1.00   1.00    .94      .98    

Not one arguable star player among Eastern teams out of the playoffs, nor from Indy or Philly. Only one such team in the West.
In both conferences, playoff teams average 52% of minutes with above-avg players; and non playoff teams avg 39%.
Cleveland used 'marginal' (or worse) players for 47% of their minutes. Miami was at .48 !
Philly, with no 'star' player, also had the fewest below-avg player-minutes (31%), followed immediately by Dallas; and no one else close.

[Crow]

This handy summary provides useful information and further analysis can yield additional important perspective.


Comparing first round winners to losers:

In the east, the winners got 3 times as many 1.5+ and 2+ performances as the losers.
Should we care more about what "star" players do against the best teams (or specific match-ups) instead of the general reliance on their average performance level?

The losers had a higher % of players over 1 on eW484 but that didn't seem to help.

Who specifically are the players between 1-1.5 eW484 and did they just hit a difficult match-up for them or have a bad stretch / bad luck or do they tend to stay stuck in this range in the playoffs and has it cost their team? Are they playing with better players or are they posting this performance level from a team role that is too big for them to produce the necessary 1.5+ or 2+ eW484? Should they be traded or not re-signed based on this analysis despite being "good" or being the team's best players?


(Can Monta Ellis at 1.41 eW484 be your #1 guy and get you to the playoffs and win something? How about Danny Granger at 1.47? Or Elton Brand at 1.43 or Iggy at 1.38? Or Nene at 1.67? Or Bogut at 1.43? It seems like a long shot. Joe Johnson as a #1 at 1.37 and the Hawks advancing can give teams hope, that if they have enough other pieces and match-up well they can do some winning in the playoffs.)


In the west the teams who advanced into the second round varied from those who did not mainly because they had about 250% as much 2+ eW484 players as the losers did. In the west, the losers almost had as many over 1.5.

So in the east the losers lacked enough stars or superstar performances but the west the losers lacked enough superstar performances.

Comparing east to west. the west leads on superstar and star performances. the east had a higher cumulative % of third tier 1+ eW484 players. The west had a higher % above 0.5 and O.

Looking at the east lottery teams, no one has even a foothold toward the apparently necessary star and superstar level of performances. In the west, everyone but Sacramento has at least a start. Not completely simple to call one team the most likely to make the playoffs and compete over the others. They all have a lot of challenge to fill and improve their star and superstar performer tiers.

(Over the last few years I was surprised by some of Otis Smith's moves and then impressed with the outcome. I was surprised by this year's moves too, but not impressed with this regular season or especially playoffs outcome. The 3 point weapon part of the inside-out game totally collapsed in the playoffs. Was it mostly a run of bad shooting luck or mostly great defensive strategy, Collins, lack of effective offensive coaching response or mostly the wrong set of complementary "stars" and role players? It is probably a mix of all, but which was the biggest part of the explanation?)


Note: eW484 does not sort players by their individual shot defense quality or impact. Other metrics might be checked to better understand that dimension.

[Mike G]

Joe Johnson as a #1 at 1.37 and the Hawks advancing can give teams hope,

Atlanta is led by Horford at 1.77 and Josh Smith, 1.74 . Both better than Nelson or any other #2 for Orlando.

I thought the Magics just shot badly, moreso than the Hawks' D was that good. Dwight abetted their cause with a lot of turnovers.
JRich, Hedo, and Arenas are clearly in decline, even as the season wore on. So were Carter and Lewis, though.

[Crow]

My mistake on Johnson. I made a quick assumption and didn't think that one thru fully. Paid like a #1. In that good but not that good slot that I was looking at. I saw nobody on that team was over 2.0 but didn't interpret the 1.5 column correctly at that moment or review the list from 1.5 - 2. I guess the same general sort of error occurred with Ellis who appears on the list above Curry and Lee but wasn't ahead on eW484. But the other 4 cases I cited were the top guy(s) on a team based on current rosters.


Will be interesting to see how drastically teams who didn't make their goals or expectations change, including those still in it.

[Statman]

Hey Mike - how do your rank players when you start getting to the below average guys? I mean - does Derek Fisher (as a current example - Bruce Bowen would be an obvious historical example) end up being one of the very worst players in the NBA this year in eWins because he played so many minutes with low production (more negative eWins?)? Is Bruce Bowen the worst guy historically in eWins?

I am certain this next question has been asked of you many times (and you've probably answered it many times - I just don't remember) - but, do your eWins tend to add up closely to actual team wins? Or, do they maybe add up to match up with Pythagorian team wins?

[Mike G]

Dan, eWins are intermediate between possession W% and pythagorean W%. That is, a team that scores 52% of the points in games (e.g., avg score 104-96) will win some 74% of it's games. The same team, winning 52% of all possessions, is assigned 62% of the eWins.
A player's eW total is just an extension of the same formulas I've developed over many years. Several production rates are scaled to team and opponent rates ('availability' of rebounds, for example). These are weighed and summed (the old T rate), and then the T rate of the hypothetical replacement player is subtracted. What's left is what contributes toward winning, and it's scaled to sum a team's eWins such that:

eW = (pW+G/2)/2 , or
pW = eW*2 - G/2

... effectively splitting the distance between .500 and game (90ish poss.) W%.
It's just our luck that the formula should be so simple; but the result is that a player's eW rate translates well from one environment to another -- like from a good team to a bad team -- if he puts up equivalent stats.

In an NBA game, on average ~484 player-minutes are recorded, with 1 win created. So 1 player eWin is average for every 484 player minutes. This season, 136 players were above 1.00 eW/484, about 4.5 per team, averaging 9092 minutes per team. That's about 46% of NBA minutes played by above-avg players.

The table above shows that as 52% for playoff teams and 39% for lottery teams.

Derek Fisher looks like a 'marginal' player (0 < e484 < .50) in eWins, at .38 eW/484 . The table shows that the Lakers needed 24% of all minutes from such players. Fish was the best of that bunch.

With 2300 minutes, Fisher amassed 1.6 eW, which was 215th in the league; typical of a 7th or 8th man. More than 300 players had fewer.
Obviously, a guy with low production who gets lots of minutes is doing something not picked up in boxscore stats; and it's probably defense. Maybe I'll find a formula that quantifies this assumption.

[Statman]

Dan, eWins are intermediate between possession W% and pythagorean W%. That is, a team that scores 52% of the points in games (e.g., avg score 104-96) will win some 74% of it's games. The same team, winning 52% of all possessions, is assigned 62% of the eWins.
A player's eW total is just an extension of the same formulas I've developed over many years. Several production rates are scaled to team and opponent rates ('availability' of rebounds, for example). These are weighed and summed (the old T rate), and then the T rate of the hypothetical replacement player is subtracted. What's left is what contributes toward winning, and it's scaled to sum a team's eWins such that:

eW = (pW+G/2)/2 , or
pW = eW*2 - G/2

... effectively splitting the distance between .500 and game (90ish poss.) W%.
It's just our luck that the formula should be so simple; but the result is that a player's eW rate translates well from one environment to another -- like from a good team to a bad team -- if he puts up equivalent stats.

In an NBA game, on average ~484 player-minutes are recorded, with 1 win created. So 1 player eWin is average for every 484 player minutes. This season, 136 players were above 1.00 eW/484, about 4.5 per team, averaging 9092 minutes per team. That's about 46% of NBA minutes played by above-avg players.

The table above shows that as 52% for playoff teams and 39% for lottery teams.

Derek Fisher looks like a 'marginal' player (0 < e484 < .50) in eWins, at .38 eW/484 . The table shows that the Lakers needed 24% of all minutes from such players. Fish was the best of that bunch.

With 2300 minutes, Fisher amassed 1.6 eW, which was 215th in the league; typical of a 7th or 8th man. More than 300 players had fewer.
Obviously, a guy with low production who gets lots of minutes is doing something not picked up in boxscore stats; and it's probably defense. Maybe I'll find a formula that quantifies this assumption.

OK, so it appears players don't get negative eWins unless they are just overwhelmingly bad statistically?

I crunched all my numbers last night - didn't have time to post anything yet - I'm guessing Derrick Fisher is around 200th in my rankings too (it's on a different computer than the one I'm on now). Our results are definitely computed & presented differently (I do that player rating that's a ratio thing - 100 league average, 85 around replacement level, players around 125 All Star level, over 140 superstar level - everything can be multiplied by minutes played and summed to perfectly match team ratings, which is simply pts_for/pts_against), but the results still are fairly similar to yours (I think my top 5 were James, Howard, Wade slightly over Rose, & Paul). From memory - I had Gasol slighly ahead of Kobe - I think I weight a little less for higher usage/lower efficiency players, and a little more for assists (and maybe steals and blocks).

Trying to make all the ratings perfectly mesh with team ratings WHILE having to deal with players playing on multiple teams is a complete pain - this is why I don't crunch my numbers that often.

I fully appreciate the fact you work your numbers as often as you do - it's always interesting.

[Mike G]

Trying to make all the ratings perfectly mesh with team ratings WHILE having to deal with players playing on multiple teams is a complete pain - this is why I don't crunch my numbers that often.
.

Crunching individual rates to create eWins compatible with pythagorean wins, this is how I get the weights, exponents, and factors that modify the raw stats. Stats correspond to wins, so it's kind of a retro-+/-, where team totals (relative to opponents) are correlated to point differential.

I call a replacement player one who produces no wins. His eW/484 = 0 .
In the league this season, I get a total of 6.1 negative eWins (from 80 players). And 1236.1 positive eW.
This amounts to about 15 players per team with positive contributions and about 3 with negative. Counting traded players separately with each team.

Not all these guys had horrible numbers. Many were just unproductive. Joel Anthony shot .565, averaged 0.8 TO/36, but just didn't do much of anything. He 'led' the league with -0.6 eW.

Of players >100 minutes, Jarron Collins (LAC) was worst at -.74 eW/484
Over 300 min., Quinton Ross of NJ, -.52
Then Anthony at -.19 for 1463 min.

[Statman]

Of players >100 minutes, Jarron Collins (LAC) was worst at -.74 eW/484
Over 300 min., Quinton Ross of NJ, -.52
Then Anthony at -.19 for 1463 min.

Over 100 minutes, looking at only per minute production (no adjustment for PT), my ratings had Jason Kapono the worst at a 15 (less than 1/6th the production of an AVERAGE player). Next worse was Luke Babbitt at 18, then Jarron Collins at 19.

Over 300 minutes, I too had Quinton Ross as easily the worst at a 31 rating.

Over 400 minutes, Stephen Graham with a 38.

Over 1000 minutes, Joel Anthony at 54.

Over 1500 minutes, Raja Bell at 56.

Those are all per minute ratings - Bell's overall rating was an 81, with the assumption that he must be doing a number of things outside the box score stats to warrant his PT. Anthony's overall rating was a 76. Graham 52. Ross 37. Since I go with the assumption that replacement player would be an 85 overall rating - all these guys are below replacement player level (Bell was fairly close).