DSMok1
Member
- Joined
- Nov 10, 2008
- Messages
- 340
- Reaction score
- 0
NCAA Players--Wins Produced
I just completed a spreadsheet analysis of every player in the NCAA, calculating their "wins produced," which means how many wins they would add to an average team playing average opponents, if they played exactly as many minutes and games as they did last year.
Some players are out of this world--Blake leads all comers with a whopping 21.773 wins produced, which is above what would be theoretically possible in such a case (an average team playing 35 games)--but he played better opposition and so could score so ridiculously high. There are only 2 other players even over 16 wins produced: DeJuan Blair and (surprise!) John Bryant of Santa Clara (who single-handedly carried Santa Clara, who had only one other player of even average caliber and a pile of players actually subtracting wins, to a .500 record and a 4th place finish in the West Coast Conference).
To make these numbers compare from team to team, I adjusted for opposition and tempo. These numbers bias somewhat toward offensive players since their stats are captured better in the box score, but I also added an adjustment for overall team defense to give a small bonus to players on teams with excellent defense. Some of that is captured with blocks and steals, but by no means everything.
These numbers are primarily for comparison between individual players rather than teams--with a little refinement, the team's players' wins produced should add up to the approximate number of actual wins (after you adjust for the opposition). The tricky part is adjusting for the opposition, so I haven't done that here. Suffice it to say--UNC and UConn have the best teams; OU's overall team was ranked 8th (about tied with Syracuse).
Here is OU's roster:
We have player, position (calculated by my spreadsheet), overall wins produced, and wins produced per 40 minutes played. Blake was most valuable, and then AJ, Taylor and Willie were second flight. A value of 0.104 WP40 rates as an "average" player at that position. This system tends to undervalue lockdown defenders, since there are no stats for that (hence Crocker being only replacement level), and overvalue players who have weak defense (since there are no stats for that either).
Even so, you can get an idea of who were the most important players on the team last year and what OU is losing--their top 3 players in wins produced.
I'm sure some would be interested in a Big XII scoreboard of wins produced, so here it is:
Notes on that list:
Here is the national top 20 list:
Notes:
The methodology of this study is based on The Wages of Wins and the position adjusted win score. I did not outright calculate the Wins Produced using his system, but rather used his PAWSmin/WP equation to approximate. I further added opposition adjustments according to my own theory to balance for opposition, which is not required in the NBA. I used average position scores I calculated from the results of my study.
Hopefully I will be able to get this spreadsheet online sometime soon for people here to mess around with.
I just completed a spreadsheet analysis of every player in the NCAA, calculating their "wins produced," which means how many wins they would add to an average team playing average opponents, if they played exactly as many minutes and games as they did last year.
Some players are out of this world--Blake leads all comers with a whopping 21.773 wins produced, which is above what would be theoretically possible in such a case (an average team playing 35 games)--but he played better opposition and so could score so ridiculously high. There are only 2 other players even over 16 wins produced: DeJuan Blair and (surprise!) John Bryant of Santa Clara (who single-handedly carried Santa Clara, who had only one other player of even average caliber and a pile of players actually subtracting wins, to a .500 record and a 4th place finish in the West Coast Conference).
To make these numbers compare from team to team, I adjusted for opposition and tempo. These numbers bias somewhat toward offensive players since their stats are captured better in the box score, but I also added an adjustment for overall team defense to give a small bonus to players on teams with excellent defense. Some of that is captured with blocks and steals, but by no means everything.
These numbers are primarily for comparison between individual players rather than teams--with a little refinement, the team's players' wins produced should add up to the approximate number of actual wins (after you adjust for the opposition). The tricky part is adjusting for the opposition, so I haven't done that here. Suffice it to say--UNC and UConn have the best teams; OU's overall team was ranked 8th (about tied with Syracuse).
Here is OU's roster:
Code:
Player Pos Win Pr. WP40
Blake Griffin PF 21.773 0.748
Austin Johnson PG 6.959 0.247
Taylor Griffin PF-SF 6.383 0.236
Willie Warren SG 6.030 0.214
Cade Davis SG 2.672 0.210
Tony Crocker SF-SG 2.432 0.094
Juan Pattillo PF 1.363 0.202
Omar Leary PG 0.883 0.103
Ray Willis SF-SG 0.571 0.222
Beau Gerber PF 0.164 0.253
Ryan Wright C-PF 0.156 0.024
Kyle Cannon SF 0.116 0.076
Orlando Allen C 0.054 0.025
T.J. Franklin SG -0.299 -0.443Even so, you can get an idea of who were the most important players on the team last year and what OU is losing--their top 3 players in wins produced.
I'm sure some would be interested in a Big XII scoreboard of wins produced, so here it is:
Code:
Player Pos Win Pr. WP40
Blake Griffin PF 21.773 0.748
Cole Aldrich C 15.456 0.596
DeMarre Carroll PF-SF 12.062 0.453
Curtis Jerrells SG-PG 11.309 0.335
Damion James PF 9.562 0.370
James Anderson SF-SG 9.519 0.325
Cory Higgins SG 9.504 0.332
Terrel Harris SG 8.863 0.319
Obi Muonelo SF-SG 8.641 0.328
LaceDarius Dunn SF-SG 8.244 0.295
Craig Brackins PF 8.082 0.311
Alan Voskuil SG 7.965 0.271
Josh Carter SF-SG 7.702 0.296
Kevin Rogers PF 7.549 0.235
Chinemelu Elonu C 7.389 0.365
Leo Lyons PF 6.979 0.324
Austin Johnson PG 6.959 0.247
Byron Eaton PG 6.721 0.242
J.T. Tiller SG 6.693 0.284
Dexter Pittman C 6.483 0.447- The top 6 players (minus one of the forwards) makes an excellent all conference squad. I'd drop James and call it good.
- Note the young players with very high WP40 rates--look for them to break out next year. In particular, watch out for Dexter Pittman as the 4th best in the conference in WP40, and second best of those returning (Aldrich the better).
- Cory Higgins carried Colorado by his lonesome.
- Voskuil was underrated
- Elonu could be very, very good.
- Sherron Collins was overrated--too inefficient with his shots. He will be the top returning point guard, though. But he was 25th best overall, and the 4th best PG this year (if you count Jerrells).
Here is the national top 20 list:
Code:
Player Pos Win Pr. WP40 School
Blake Griffin PF 21.662 0.744 Oklahoma
DeJuan Blair PF 19.467 0.815 Pittsburgh
John Bryant C 17.112 0.669 Santa Clara
Terrence Williams SF-SG 15.850 0.501 Louisville
Hasheem Thabeet C 15.792 0.552 Connecticut
Cole Aldrich C 15.375 0.593 Kansas
Kenneth Faried PF 15.153 0.560 Morehead St.
Ahmad Nivins PF 14.786 0.470 Saint Joseph's
James Harden SG 14.065 0.449 Arizona St.
Jeff Pendergraph PF 14.011 0.492 Arizona St.
Luke Harangody PF 13.637 0.469 Notre Dame
Taj Gibson PF 13.438 0.456 Southern California
Manny Harris SG 13.195 0.458 Michigan
Ty Lawson PG 13.158 0.502 North Carolina
Trevor Booker PF 12.949 0.527 Clemson
DeMarre Carroll SF 12.930 0.486 Missouri
Aaron Jackson SG-PG 12.806 0.410 Duquesne
Jordan Hill PF 12.679 0.418 Arizona
Tony Gaffney C-PF 12.460 0.490 Massachusetts
Patrick Patterson PF 12.424 0.433 Kentucky- Thabeet, much maligned, was still rather good.
- A lack of top-flight guards this year--Harden at SG and Lawson at PG are the best.
- Manny Harris sure didn't look that good against OU and Tony Crocker!
The methodology of this study is based on The Wages of Wins and the position adjusted win score. I did not outright calculate the Wins Produced using his system, but rather used his PAWSmin/WP equation to approximate. I further added opposition adjustments according to my own theory to balance for opposition, which is not required in the NBA. I used average position scores I calculated from the results of my study.
Hopefully I will be able to get this spreadsheet online sometime soon for people here to mess around with.
Last edited:
