How Good is Kevin Durant and the Thunder?

Posted on November 25, 2009 by


Last October this was quite the debate.  Just to recap:

  • The media and coaches have always liked Durant.  He was a top pick in the draft and voted to the All-Rookie team by the coaches.  And the media said he was Rookie of the Year in 2008.
  • John Hollinger’s Player Efficiency Rating has also always said Durant was “good”.  Both PERs and the media are heavily influenced by a player’s point totals, so it’s not surprising to see PERs and the media agree (and the same can be said for coaches).
  • Wins Produced said Durant was below average as a rookie, but improved dramatically in his second season.  How did he improve?  His shooting efficiency, rebounding, and steals went up while his turnovers went down.  In other words, Durant improved with respect to the primary box score statistics that increase wins.
  • Despite the improved box score numbers, though, adjusted plus-minus insisted that Durant was below average in both his first two seasons.  The difficulty with this measure is that it never tells you why a player is “good” or “bad”.  But since Durant had very good box score numbers, it was argued that Durant was actually very bad at on-the-ball defense.

Of course, there was another potential explanation.  Adjusted plus-minus is a very inconsistent measure.  The year-to-year correlations are very low (there is a great deal of “noise” in the model).  So it could be Durant’s adjusted plus-minus was due to his supposedly awful defensive skills.  Or it could be noise in the model.

After 15 games this year it is beginning to look like “noise” was the culprit.  According to, Durant’s adjusted plus-minus score is 16.09.  This is the second highest mark on the team.  So it looks like everyone is now in agreement.  Durant is officially a “good” player (ain’t that a relief?).

Fortunately for the Thunder, he is also not alone.  Here are the top producers of wins in Oklahoma City [WP48 = Wins Produced per 48 minutes].

Kevin Durant: 579 minutes played, 0.204 WP48, 13.5 projected Wins Produced

Thabo Sefolosha: 491 minutes played, 0.182 WP48, 10.2 projected Wins Produced

Russell Westbrook: 512 minutes played, 0.138 WP48, 8.0 projected Wins Produced

James Harden: 296 minutes played, 0.219 WP48, 7.4 projected Wins Produced

This quartet are on pace to produce 39.0 wins.  As a team, Oklahoma City has an efficiency differential (offensive efficiency minus defensive efficiency) of 2.4, a mark that projects to about 47 wins across an 82 game season.  So the four players listed above are responsible for about 83% of the team’s wins.

Missing from this list is any player at power forward or center.  Jeff Green remains a below average player.  And Nick Collison and Etan Thomas have yet to produce much in 2009-10.  D.J. White, though, has posted a 0.423 WP48.  Before Thunder fans get too excited, White has only played 44 minutes and has yet to play more than 15 minutes in a game (or more than two games in a row). 

Despite the problems in the frontcourt, it does look like Durant is a “good” player.  And because he has three other “good” teammates, it looks like the Thunder will be “good” this year.

Let me close by noting the play of James Harden. So far Harden is only averaging 19.7 minutes per game.  But if his WP48 of 0.219 continues – and his minutes per game don’t change – he will produce 7.4 wins by the time the season ends.  Such production could be what pushes the Thunder past the 41 win mark; or the mark that divides “good” teams from “bad” teams.  Harden probably may not play enough minutes to warrant consideration for Rookie of the Year (he plays the same as Thabo Sefolosha).  But in a discussion of “good” rookies in 2009-10, Harden’s name should be mentioned.

– DJ

The WoW Journal Comments Policy

Our research on the NBA was summarized HERE.

The Technical Notes at provides substantially more information on the published research behind Wins Produced and Win Score

Wins Produced, Win Score, and PAWSmin are also discussed in the following posts:

Simple Models of Player Performance

Wins Produced vs. Win Score

What Wins Produced Says and What It Does Not Say

Introducing PAWSmin — and a Defense of Box Score Statistics

Finally, A Guide to Evaluating Models contains useful hints on how to interpret and evaluate statistical models.