Mere Addition Paradox

I was looking at ESPN.com's "Hollinger stats" page and was surprised to see that in terms of Hollinger's PER stat, Durant is the 26th best small forward in the league. Not great, by any means, but good enough to be a starter somewhere. And, indeed, by the formula Durant is just ever so slightly below average.

I'd been under the impression that Durant was actually playing terribly. So I looked up the breakdown. It seems that in terms of scoring efficiency, Durant is pretty bad -- 50th among small forwards in true shooting percentage. As a rebounder he's worse -- 55th best rate among small forwards in rebound rate. 52nd in turnover ratio, and 53rd in assist ratio. Basically, he seems to be a bit worse than the fiftieth-best small forward in the league. That's probably good enough to get some minutes as a backup, especially since in light of his age he may well improve if he gets a chance, but it's a far cry from 26th best as Hollinger's aggregate statistic makes him out to be. What accounts for the difference?

Well, it turns out that Durant does excel at one thing -- getting plays called for him. He's got the third-highest usage rate among small forwards. But does he really deserve to get the level of credit for this that Hollinger's giving him? I mean, if you're not a very good player, your usage rate ought to be low. Using tons of possessions isn't helpful if you're at Durant-like levels of effectiveness.

Comments

Yeah, but when you're discussing a player like Durant, its all pretty moot, anyways. He's a rookie who's been asked to carry the offensive load for a terrible team. Regardless of his PER, (which I actually hold in high regard)you would not take 50, or evn 25 small forwards ahead of him. Imagine if he was coming off the bench for a team like the Pistons for example. He would be a MUCH more effective player and his PER would certainly reflect that. I see your point, Matt, but I think Durant is doing just fine, all things considered.

Posted by Todd Doug | March 5, 2008 8:32 PM

Durant's been playing shooting guard for a while now.

Posted by Jeremiah | March 5, 2008 9:42 PM

He should be a 3 or 4, I think. But being 50th best at a lot of things may actually make him 26th, regardless. Also, I'm not sure that PER includes usage rate as a component.

Posted by RM | March 5, 2008 10:34 PM

My guess is that usage is weighted the way it is because the more touches he gets, the more predictable the offense becomes, the easier it is to defend against, the harder it is to score. So it's the classic Pareto-style multi-criteria optimization problem, where you have two competing objectives (usage and everything else) and where Durant simply falls way past the point on the usage scale where his numbers on everything else have pushed the Pareto frontier to the point where he looks terrible on paper offensively.

Posted by msmackle | March 5, 2008 11:46 PM

If you look carefully, you'll see that it's actually the little-known small forward Kevin Druant who Hollinger's has ranked 26th best. The sucky but oddly popular Kevin Durant is not discussed.

Posted by southpaw | March 6, 2008 12:34 AM

What Todd says is true. In addition, remember that the Sonics are a team with an owner that is most concerned right now with putting his team in cold storage so that he can they can be moved out of Seattle. The owner has no incentive to improve the team and the players have no incentive to work together or improve.

Durant is the real deal, I think. He's just in a crappy situation.

Plus, PJ is a crappy coach.

Posted by stand | March 6, 2008 1:41 AM

This is the main problem with PER. The more shots you take, the higher your PER, with almost no regard to how efficient you are. It works fine for comparing efficient shooters, but that's why players like AI, Durant and others are WAY overrated by it.

Although to be fair, AI is having one of his better seasons. That's not to say that Durant could be good in the future, but the point of stats is to assess how good someone is performing right now, and in recent or older history. And Durant is just terrible, and torpedoing his team out of games.

Anyway, the wages of wins critique of PER is more sophisticated than mine. If you want more reading on the subject.

Posted by Rashad | March 6, 2008 9:22 AM

Why do these posts sometimes show up, like, hours after the stated time? It's really odd.

And, also, what Rashad said. To me, this is the biggest problem with PER. Meanwhile, anybody notice that all of a sudden Carl Landry of the Rockets is showing up on these statistical measures as by far the best rookie?

Posted by Al | March 6, 2008 10:15 AM

Carl Landry has had an amazing year. Either that, or its a very strange coincidence that the Rockets started winning lots of games when he started playing.

Matt - You being a bit coy here? Durant is being rewarded by PER for the shots he takes, rather than punished for his below league average scoring efficiency. That's the WOW critique, and it certainly is correct in this case. The reality is, there are very few players in the league who have done more to hurt their teams winning percentage this year. Things may change next year. But it would be a total crock if he won ROY.

Posted by Flint | March 6, 2008 11:41 AM

Reminds me of Antoine Walker. He was an awful shooter who killed whichever team he was on. But he claimed to be a "volume shooter" who could score a lot of points if he took a lot of shots. Huh? Everyone can score a lot of points if they just take enough shots.

Posted by Eric | March 6, 2008 11:53 AM