[nba.com] League's best one-on-one defender with statstical evidence

I dont think there's ever been any real dispute about who's the more talented player based on overall skills/abilities. It's been about who fit better with Houston.

 

How is Ariza as a one on one defender? How does he compare to Ron Artest and Battier.

 

Gay fits way better with Houston.

 

I already said it classifies the player correctly 75% of the time. You know that in advance, hypothetically.

Would it be useful?

 

it would also be wrong, as the NBA inflates the value of offensive production and deflates the value of defensive production by the very nature of the way the game is measured.

good offensive players that are bad defensive players almost always make more than the inverse. with a few notable exceptions.

so its not a fair market.

 

Not if that 25% is very incorrect.

 

I found myself asking the same question. I would love to see where he rates. I would also like to see the updated numbers for this season.

 

i wrote "fit" not "fits" but your answer may still be the same

 

i think durvasa is trying to tell you that if you know it's always correct 75%, then you can say it's pretty reliable because thats a pretty high percentage, so therefore it's useful in figuring somethng out.

I'ts like having a machine that diagnoses car trouble, but only get its right 75% of the time, it may not pinpoint your problem, but there's a high chance it will, if it gives you expanded possibilities as well ranked based on their probability, then it would, at the very least, help you narrow down the possible problems, so that becomes a useful tool to have even if it's not 100% right.

 

If that's the case, then clearly Battier was better at the time we made the trade? Because he made more money in the open market.

In some sense it doesn't even matter if Battier is "better" anymore - we got our 3 years of good production out of him as a solid fit for this team, and could always try and re-sign Gay if we wanted anyway.

 

In "moneyball" terms, Durant is basically a poor-man's Rudy Gay (not so much this year, moreso in the past couple of years).

 

That assumes perfect information shared by all parties. As Morey has stated time and again, Houston believes it has better (proprietary) information that most other teams in the league do not.

In real life, capitalism would indicate the best teams pay the least for the best talent, the worst teams pay the most for mediocre talent. Kind of like the stock market... it's no coincidence that Les made hundreds of millions in derivatives under "capitalism" while regular shmoes watched their 401ks implode the past year

2 things about the article. raw On-Off Court Differential doesn't necessarily correlate to "efficiency" since it's much harder to raise the Differential if you're on a good defensive team. You can see the top of that list are dudes from teams that are NOT elite defenses, quite the opposite

second, does this guy watch the Rockets? way too much talk about Artest and Battier as if those are the only 2 guys on the team who play defense. Although Artest and Battier did play together on some rotations, there's going to a much higher correlation with interior defenders (and good defenders) that Houston had last year a la Chuck Hayes, Deke and of course Yao.

 

great post. props.

 

I'll try again.

We have a statistical model that classifies a player in one of three categories:

(1) very good defender
(2) very bad defender
(3) somewhere in between

Through extensive cross-testing with experts (coaches/scouts, for instance), we determine that this model correctly classifies a player ~75% of the time. The rest of the 25% of the time, it misses.

Is this useful?

I would say yes. If it happens to miss on some players (e.g., Nick Young is a very good defender), that does not mean we should throw it out.

 

If that were true then teams would more premium on defense. That is capitalism.

 

This isn't really an argument for one side or the other, but what's better:

- A stat that's 100% right 75% of the time
- A stat that's 75% right 100% of the time

 

Kevin Garnett deserved every bit of that DPOY trophy...

 

I know what he's trying to say. I just don't agree, at least in terms of basketball. The point of these stats are to prove, objectively, how good a defender someone is. When I look at these stats they're either telling me something I already know or worse, telling me something I know is wrong.

Your analogy for a car diagnostic doesn't fit in with basketball imo because a) while having a possibility of being wrong they aren't way off. You won't get a reading that says you have a flat tire when you need a new transmission. b) It's a car. If you're wrong, you move on and try something else. The price is insignificant compared to professional sports.

A better analogy would have been a doctor running tests on your body. The machine may give a false reading and you might be misdiagnosed but it won't be way off. If you're have a brain tumor the CT scan won't show that you need to amputate you have a broken foot.

My gripe isn't with the stat being wrong in some instances it's with how wrong it is in those instances.

 

defense is much harder to quantify and therefore much harder to compensate someone for.

and no, that's not capitalism.