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

i think you're focusing too much on right vs wrong, durvasa was just pointing out that at the very least it's useful. It's hard to say with 100% certainty who the best defender is, no one's formula will ever really give you that, but if you have a group of formula's who are each reliable most of the time, like 75% of it, then they can be used to get a general idea. You can't keep throwing them out looking for a perfect one.

 

This depends on how much you know about the remaining 25% (i.e. is there any way to figure out which cases the 100% right model fails on -- though in most practical situations the answer is no. Given that, I would prefer the 75% right 100% of the time stat, especially if there is high variance).

Anyway, of course the stats presented in the article are useful. Props to durvasa for making this point in a non-confrontational way.

IMO, the defensive info on Rudy Gay is just part of a larger trend with evaluating him -- he does well on raw traditional stats but poorly on stats that include some sort of calibration for expectations. In fact, there are a lot of players who could put up stats similar to his. Just look at Ariza. Even without having star-like offensive skills Ariza is putting up offensive numbers comparable to Gay now that he's moved to a team without an allstar. And Ariza plays much better defense.

 

No.

I think something that is right 60% of the time but is close to being right that other 40% of the time is much more useful. You are speaking hypothetically so if I'm a professor and a grad student hands me a test that is 75% right but that other 25% is way off he's showing me he doesn't know what the heck he's doing. I would not trust him to work with me, he is not useful because I can not trust his future work.

Even in your hypothetical situation where we are somehow certain this model is right 75% of the time it is almost disingenuous to use it in any objective way (which I thought was the goal here) when the other 25%'s correctness varies so wildly.

 

that's not a scientific way to accept a model.

statistically, you still have to do a chi-square test to decide the model useful or not, which means those 25% misses do count.

 

listen bob battier is a top 5 defender while ariza is a defensive talent like a verry verry poor mans ron artest if u watched the phoenix game u saw that arizas man would back cut him for easy lay ups and ariza would die on the screens as his man drained an open shot u can talk about how ariza has better potential for the future but right now battier is better than that man and its not even close were a poor defensive team cuz arizas not doing his part defensively hes to focused on the offensive end and its taking away from his defense ur stupid.

 

agree with this.

if it misses i'd prefer it to be close. its the hand grenade theory of life.

any metric that can be horribly wrong destroys your confidence in it, even when it's right.

 

Thank me later.

 

but your answer is only "no" because you're comparing the 75/25 to your example of 60/40. the question is just, is the 75/25 useful or not? Not is it useful compared to your 60/40...

 

Spoiler

Here are their stats for this year.

Ariza's averaging a couple more assists, but Gay turns the ball over less and scores more on fewer shots. In terms of efficiency, it's not even close to being close.

- 20.5 points/game, .563 TS%, 11.1 TOV%
- 18.4 points/game, .503 TS%, 13.4 TOV%

Unless the second guy is racking up Nash-like assist numbers, I don't see how you can say those two players are performing comparably on offense.

Also, if you're into the more advanced stats, Gay's plus/minus (both adjusted and unadjusted) is better than Ariza's, and he has 0.8 offensive win shares to Arizas 0.2.

If Ariza's offensive numbers were comparable to Gay's, we'd probably be undefeated right now...

 

Caption this:

 

I really hate these statistics...

Just because a player leaves the court doesn't mean the team will just implode defensively. Just because a player enters the court doesn't mean the team will become a better defensive unit.

They judge these on the basis that everyone has consistent defense, all the time. Also, some teams play more help defense, than other teams do. Therefore, these statistics are way too inaccurate to be measured.

 

I understand what you're saying, but if kobe is shooting 47%, then that's very,very efficient as a scorer. It doesn't matter nor does it determines the outcome as much as what other will do. The rockets beat la 3 times mainly because they have no one on their roster that can contain brooks. Its not like kobe was shooting sub 35% like he did in the finals vs the pistons or vs boston. The goal is to limit everything and or force a poor %, but with great scorers, its more of when they score. If kobe is struggling through 3 qts and the last qt he scores 14 pts on 5-7 shots, but for the game he was 10-22, then he wasn't contained.

I think shane does a good job, just not the shutdown stuff people like to say he does on this board.

 

And yet Morey and Adelman have Battier as the team captain and vocal leader? Gay is talented, much more on the offensive end than Battier, but his defense simply does not exist. He wouldn't fit in here at all. We don't need a volume shooter with no defensive abilities.

 

Well my point is not that TA is better than RG, but that he's comparable. First consider some stats before this year (which has a small sample size).

Rudy Gay's career TS%: .530
Trevor Ariza's career TS%: .528
Pretty close wouldn't you say?

I looked up their OWS from last year and I was surprised to see that TA's OWS was actually much higher: RG 0.7, TA 2.5.

Yes, Gay's shooting has been good this year and Ariza has been bad, though I suspect that's the small sample size effect. I don't have the data to prove this, but I would guess that assists are a less noisy stat (i.e. your A/TO ratio is more stable from 1 week to the next than your shooting %). And the difference in assists is frankly huge. Ariza is averaging 3.9 A/2.7 TO. Gay is averaging 1.7 A/2.1 TO. Essentially every top wing player in the league has an A/TO ratio > 1, and any wing player with A/TO < 1 is at best mediocre. I would be impressed if you can find data to suggest otherwise.

 

The extra hand helps but it is not the single reason why Kobe is high.

 

But, shooting 47% while limiting free throws and easy looks for teamates is what wins games for the Rockets.

This is a great thread BTW, but I am sure the anti-Battier brigade will not be swayed.

 

For the record, as it stands Gay is in career-best shooting form @ 48% from the field.

When you "pace adjust" their stats per 40 we find:

Gay is on 19.9ppg
Ariza on 18.6ppg (pretty darn close)

But the killer is Ariza's 4.2apg compared to Gays 1.7 - that's effectively the creation of another 5ppg (ie - 2.5 extra apg) - or perhaps a tad more when you factor in 3pointers etc.

Again, the pace adjusted turnovers aren't much - 2.2 for Gay, 2.8 for Ariza - which is probably in Ariza's favour considering he has the better passing insticts to the open man for the bucket.

Oh - throw in the fact that Ariza has better range (41% from 3 compared to 35%) and there's not too much between them.

Sure, Gay gets to the FT line more often (which skews his TS% to be slightly higher) - but there isn't too much between them offensively based on THIS SEASON.

It's a tough argument to make that "Gay is waay better offensively" based on what they are producing THIS SEASON.

 

Pretty close, except Gay's career USG% is 24.2, while Ariza's is 17.7. All else being equal, as USG% increases, TS% decreases, because that extra usage comes in the form of lower-percentage shots than the shots you were already taking.

I say "all else being equal" because there are plenty of examples of players improving their TS% while increasing their USG%. Actually, this is the case for many players, if not most. But those players' usage increased because their TS% increased. In other words, they started taking more shots because they improved their game.

Could Ariza become a more efficient shooter from here on out? Sure. But if you took the Ariza of the past 5+ seasons, the one who had a TS% of .528, and raised his USG% by 35%, logic suggests that it would cause his TS% to go down.

Well, I should state that I'm not that big a fan of Win Shares as a statistic, so I'd be fine with just disregarding it and using other measures instead. But for what it's worth, WS are largely based on your team's record, which makes it even more remarkable that Gay has more WS than Ariza this year. Obviously the formula is way too complicated to go into in detail, and I'm far from an expert, but it's something along the lines of:

WS = Number of wins by your team x (efficiency x production)

Basically, a teams' wins are divided amongst it's players, with the most efficient and productive players getting the most shares. Like I said, not my favorite metric, but interesting to look at nonetheless (especially when a player on a bad team has more WS than a player on a good team).

Carmelo Anthony

Kevin Durant

Danny Granger

All three have a career A/TO ratio lower than 1. And come to think of it, Gay's numbers are probably more comparable to those three, than Ariza's numbers are comparable to Gay's.

 

I'm not quite sure what you're saying. If it misses 25% of the time, that's of course important to know. But that doesn't change the fact that a model that correctly classifies a player's defensive ability the majority of the time is useful.

Think of it as an indicator. Suppose I know that 75% of the time that my dog barks for no apparent reason at 5pm, it starts raining 15 minutes later. Suppose I need to walk to a store to meet someone at 5:30, and the question is whether I should I bring an umbrella. I can't look outside, and I can't check the weather on TV/online. Whatever decision I make before I open the door is what I'm stuck with.

Would it be useful to know whether or not my dog barked at 5pm? I think yes.