PDA

View Full Version : top 10 players in various advanced stats



PatsSoxKnicks
06-15-2011, 06:11 PM
All of these advanced stats are "all in one" metrics and they are based on the regular season only. I'll provide a link for some of the lesser known stats (such as WARP).

I'll be posting them over a couple of days/weeks. At the end, I'll do a "composite top 10". I'll probably be posting the +/- numbers later (I want to split adjusted +/- into offense and defense)

Here are the top 10 players in Win Shares (http://www.basketball-reference.com/about/ws.html):


Player Team Off WS Def WS WS
LeBron MIA 10.3 5.3 15.6
Pau G LAL 10.0 4.7 14.7
Dwight ORL 6.7 7.7 14.4
CP3 NOH 9.5 4.4 13.9
D Rose CHI 8.4 4.8 13.1
D Wade MIA 8.1 4.7 12.8
Durant OKC 8.6 3.3 12.0
Pierce BOS 6.5 5.1 11.6
K Love MIN 9.0 2.5 11.4
Dirk N DAL 7.8 3.3 11.1


Obviously, the most well known advanced stat and can be found on basketball-reference.

The top 10 players in EWA (http://sports.espn.go.com/nba/columns/story?columnist=hollinger_john&page=PERDiem-090325) (Estimated Wins Added):


PLAYER Team PER VA EWA
LeBron MIA 27.34 770.0 25.7
Dwight ORL 26.13 680.3 22.7
D Wade MIA 25.65 638.3 21.3
Durant OKC 23.70 598.4 19.9
D Rose CHI 23.62 569.8 19.0
Kobe LAL 23.94 557.4 18.6
CP3 NO 23.76 548.4 18.3
Russel OKC 23.63 536.8 17.9
Pau G LAL 23.33 536.1 17.9
K Love MIN 24.39 502.4 16.7


EWA is simply converting PER into how many wins you added for your team.

The top 10 players in WARP2 (http://sonicscentral.com/warp.html):


Player Tm Ortg DRtg Win% MP WARP
Lebron MIA 111.8 103.6 0.744 3063 21.0
Dwight ORL 108.6 100.3 0.751 2935 20.5
D Wade MIA 110.4 103.7 0.706 2824 17.1
CP3 NOH 110.4 103.9 0.699 2880 17.0
D Rose CHI 110.6 104.8 0.680 3026 16.7
K Love MIN 111.0 103.9 0.716 2611 16.4
Pau G LAL 108.7 103.6 0.662 3037 15.6
Durant OKC 108.8 104.5 0.637 3038 14.1
Russel OKC 109.6 105.1 0.641 2847 13.4
Manu SAS 110.2 104.5 0.677 2426 13.2


WARP is one of the lesser known metrics out there. Basketball Prospectus recently came out with WARP2 which is explained here (http://www.basketballprospectus.com/article.php?articleid=1209).

On a side note, Basketball Prospectus published their top 10 in WARP2 in their MVP article (http://www.basketballprospectus.com/article.php?articleid=1633) which came out on April 8th. Obviously, that was before the season finished up. However, I projected out the top 10 totals using their final minutes played total on the season and their Win% from the MVP article. So the assumption being made here is that each player played at the same level in the last week as they did before April 8th. Obviously, not that big of a deal so these numbers should suffice.

The top 10 in Wins Produced (http://www.wagesofwins.com/CalculatingWinsProduced.html):


Player Team Pos Adj P48 WP48 WP
K Love MIN 4.22 0.760 0.474 25.8
Dwight ORL 5.00 0.716 0.382 23.4
LeBron MIA 3.19 0.554 0.356 22.7
CP3 NOH 1.00 0.521 0.358 21.4
D Wade MIA 2.00 0.450 0.322 18.9
Zach R MEM 4.31 0.583 0.291 16.5
Pau G LAL 5.00 0.578 0.258 16.3
Blake G LAC 4.30 0.538 0.248 16.1
KG BOS 4.00 0.595 0.323 15.0
K Hump NJ 4.00 0.616 0.344 14.8


Pos is short for Position.

I'll admit, I'm not a fan of this metric at all. K Hump is short for Kris Humphries. Yes, you read that right. Kris Humphries is in the top 10 for Wins Produced. Berri, the creator of Wins Produced, seems to have alienated himself from the rest of the APBR community. His metric has been criticized for valuing defensive rebounding too much and basically ignoring the Usage vs. Efficiency tradeoff. Nonetheless, I decided to include this for you guys to see.

The top 10 in eWins (http://sonicscentral.com/apbrmetrics/viewtopic.php?f=2&t=211&sid=454934b03b927b44088143c1c9d5183c):


Player Team Eff% e484 eWins
Lebron Mia 0.581 2.79 17.7
Dwight Orl 0.594 2.57 15.6
D Rose Chi 0.54 2.45 15.4
D Wade Mia 0.568 2.46 14.3
Kobe LAL 0.538 2.43 14.0
Blake LAC 0.535 2.11 13.5
Durant OKC 0.576 2.13 13.4
Russel OKC 0.526 2.22 13.1
Pau G LAL 0.577 2.04 12.8
CP3 NOH 0.566 2.04 12.1


From wikipedia:


Mike Goodman is the creator of the EWins method for evaluating the importance of player contributions to team winning. It makes use of its own custom weights & unique adjustments and presents player value in terms of wins added (per season or playoff series).

He posts his numbers regularly on the APBRmetrics forum. For an explanation on his method, here are 3 (http://sonicscentral.com/apbrmetrics/viewtopic.php?f=2&t=162&sid=32902633070e82b7a3add4274a94d5de) different (http://sonicscentral.com/apbrmetrics/viewtopic.php?f=2&t=209&sid=32902633070e82b7a3add4274a94d5de) threads (http://sonicscentral.com/apbrmetrics/viewtopic.php?f=2&t=70&sid=32902633070e82b7a3add4274a94d5de) that might help.

I'll be posting the +/- numbers (adjusted, statistical, RAPM, etc.) eventually. I'm planning on converting them into "win" type metrics so the numbers are all on the same scale (I'll post the raw numbers too). I also want to split adjuted +/- into offense and defense which might take awhile. After that, I'll do a "composite top 10".

Anyways, post your thoughts etc. on which metric you like best.

Chronz
06-17-2011, 02:08 AM
I was thinking about doing one of these for synergy, just to have an archive once it inevitably gos private.

PatsSoxKnicks
06-17-2011, 02:15 AM
I was thinking about doing one of these for synergy, just to have an archive once it inevitably gos private.

lol I was thinking of doing the exact same thing. I put all of this years numbers in a spreadsheet for the all-defensive team caliber players. I was planning on posting those numbers along with some of the other defensive stats (like 82games.com counterpart PER, on/off court DRtg) at some point.

And I'm assuming you're talking about the free data from my synergy sports?

Patman
06-20-2011, 05:48 PM
Interesting Collection, most of the Metrics agree more or less about the top Players. All those Stats that try to get the quality of a Player in one number have their flaws, but they are good indicators about who the best players are.

I really don't like Berri's Wins Produced it has way to many outliers for it to be taken seriously.

I would really be intrested in the Stat or better Chart Dean Oliver had worked on for players, where he mapped USG% to ORTG, in his book he has some examples, but i think he wasn't finished with it. But if find this one of the more important topics because it shows that it is hard to keep the efficency up when the USG% nears 30.

PatsSoxKnicks
06-21-2011, 02:47 AM
Interesting Collection, most of the Metrics agree more or less about the top Players. All those Stats that try to get the quality of a Player in one number have their flaws, but they are good indicators about who the best players are.

I really don't like Berri's Wins Produced it has way to many outliers for it to be taken seriously.

Yeah, I think Berri's Wins Produced is a joke. I was debating whether to even include it. Kevin Love is a very good player but there is no way he should come out as the most valuable player/best in the league. And Kris Humphries at 10? lol.

Most of the other APBRmetricians don't take Berri seriously. And as you mentioned, his Wins Produced metric has way to many outliers to be taken seriously by anybody.

I'm not sure if I'm going to include his metric in the "composite top 10" that I plan on doing when I'm done posting the +/- metrics.



I would really be intrested in the Stat or better Chart Dean Oliver had worked on for players, where he mapped USG% to ORTG, in his book he has some examples, but i think he wasn't finished with it. But if find this one of the more important topics because it shows that it is hard to keep the efficency up when the USG% nears 30.

This would be interesting to see. A scatter plot graph in excel could probably illustrate the relationship nicely. This would be easy to do too.

Patman
06-21-2011, 01:41 PM
This would be interesting to see. A scatter plot graph in excel could probably illustrate the relationship nicely. This would be easy to do too.

Yeah if you only wan't to do a correlation over multiple players it isn't that hard, maybe will make something like that after my exams are through.

But if you are interested how a specific players do at different USG% it get's harder because you have to parse single games to see how they do when they have high USG%, when it is low and so on. Then you have to take care of outliers and account for different Team setups etc. After that you have to extrapolate the whole thing for USG% you do not have big enough sample sizes. It get's complicated and i honestly don't have the statistical knowledge, time and data to do something like that.

But in Olivers Book you see that a absolutely top player like Jordan could hold his ORTG extremely long before he had an significant drop of.