OKC Trade/Signing Deandre Jordan, Reggie Williams

Let’s not forget the defense, angle, too. I haven’t seen Jordan enough to talk about his defensive awareness, but he has the physical tools to be a pretty outstanding defender and I think his reputation is already as a solid one? As in, better than Biedrins. Plus he has excellent physical tools – a very good combination of height, weight/strength, length and athleticism, which gives him a high ceiling.

Right now there’s an outstanding bid of 22 million for 4 years for Perkins, does anyone think that will double due to a bidding war? I’d be equally happy with either of them.

Yes, I do. That offer is essentially meaningless. It was the starting point in the negotiation from a team who, at the moment, has his exclusive rights. Out of curiosity, do you have a link to the $22M number?

Jordan is in the up and coming category that would suggest he will get overpaid. But I think 11 million for him is a joke.

Trust me, I see this logic, too. I just stopped accepting it when I took some time to look through various C’s in the league and find out what they’re being paid. Bynum’s making ~$14M this year, and that number is going up over the next couple seasons. Tyson Chandler is making almost $13M. Rasheed Wallace, right now, is being paid just $13M for this year and next. Brendan Haywood has 5 years, ~$42M starting this year, with a team option over $10M after the 5th year. Jermaine freaking O’Neal signed for 2 years, ~$12M. Kenyon Martin is making ~$16.5M, Nene $11.3M. Al Harrington has 5 years for over $33M. Al Jefferson has 3 years for $14M average. Okur $9.5M this year, $10.9M next. Yao $17.7M. Scola 5 years for $47M. Al Horford has a 5 year/$60M extension starting next season. Josh Smith 3 years/$37M. Bogut 4 years/$50.6M. 5 years, $32M for Gooden. Elton Brand is getting about $17M average for 3 years.

I compiled this list merely by looking at 10 teams (though in the interest of full disclosure, they were the top 10 teams in salary according to Hoopshype). It’s not meant to be representative, merely to illustrate that there are a lot of really tall guys in the NBA being paid a lot of money, many of them far more than they’re worth. Big men get paid a lot. It doesn’t make sense how much they paid, but it’s reality.

Jordan was pretty good last year, and really isn’t producing all that differently this year. If Griffin is having any impact on his play, it doesn’t look substantial to me. Where it might be showing up, in my opinion, is in a lesser offensive role for Jordan (scoring less, turning it over less), which has made him a bit more efficient with both scoring efficiency and less turnovers but also decreasing his scoring rate, and possibly in taking rebounds away from him.

So when we say regression towards the mean, we are actually talking about a mean that is not fixed, and changes over time (drops every season).

Actually the mean is a mythical, unknowable number. I’m actually being serious here. The way we think about it statistically is a player has a “true talent level” – basically, given a sample size of infinity, it’s how he’ll play. It is exactly how good he is. Only we don’t know what that number is, and we’re merely sampling to try to estimate it. And that mythical number is constantly changing, though the nice thing is from a statistical standpoint, we’re best off assuming it’s either static, or gradually trending in the observed direction or in the direction the population tends to move over time (from the point of view where we’re trying to estimate it based on our sample, this will be the most accurate way of estimating it). For instance, we know players in their 30’s tend to get worse every year overall. A player in their 30’s might be better one day, worse another, and might even be better for a whole year than they were the previous year, but we’ll be most accurate on the whole assuming they’re worse this year than they were last year.

The thing is, if he shoots less FTA/year going forward, his poor ft% will have less of an impact on his overall ft% (mean), because it will be based on less FTA (smaller sample size than prior to 09-10) so he wont actually regress towards that career mean (which is not fixed and is moving down, but at a slow rate). For him to regress toward the mean, he would have to A) get his ft% back up, or B) get his FTA’s up so that his current, lower FT% has an equal impact on his mean FT% as his previous, higher FT% did.

The way regression towards the mean works is that a player will play at his talent level going forward, on average. So if Biedrins is truly a 55% FT shooter, he will shoot 55% going forward, on average (even if he’s shot worse than that so far). As our sample size increases, our sample mean will move closer to 55% – it’s a limit of 55% as n goes to infinite. If he doesn’t get to the line often, that means it will take a long time for his FT% to get close to 55%, and if you cut it off at a season, it probably won’t even get that close. The important thing going forward, though, in this example where Biedrins is a 55% FT shooter is that’s what we expect him to shoot in the future.

With Biedrins specifically, if we’re trying to estimate his talent level for FT%, we have to weight his performances by sample size. This year and last year get very little weighting, since he only shot 25 each year. Previous years he has a lot more, so they mean a lot more, which is why we should be very confident he’ll shoot much better going forward. Guys just don’t all of a sudden forget how to shoot FT’s. It’s possible this somehow happened to Biedrins (and even more likely it happened on some scale, even if it’s a small effect), but it’s much more likely random variation is the biggest factor at play, so that’s what we should assume.

Like I said before, I think we have information to conclude with pretty good confidence Biedrins is not getting to the line very well, and his current “true talent level” is much lower than it used to be. I don’t see evidence for the same in FT%. Sample size is just so small. The best conclusion, in my informed opinion, is that he will regress towards the mean in FT% (whatever that mean is, it’s definitely something much, much higher than we’ve seen lately), but like I said, that means shooting his mythical “true talent level” going forward, which doesn’t necessarily mean his actual stats for this year will ever get back to that mean.

You’re obviously very encouraged by Ab’s more recent results and better form.

I suppose, but actually that has nothing to do with my reasoning. My reasoning is entirely sample sized based and an understanding of the potential of random variation.

But asserting that he will regress to the mean completely ignores human psychology.

Actually, it doesn’t ignore it at all! Not even a little bit! It’s great because both things exist together in reality. All regression towards the mean is, is a statistical phenomenon that comes about from random variation, which exists in pretty much every setting you can possibly think of to take a measurement of anything. Stock prices regress towards the mean, the effect of gravity will regress towards the mean, all sports statistics will regress towards the mean. In a sense, regression towards the mean, as a concept, is admitting we don’t actually know exactly what’s going on, and simply taking the best odds (which we get from sampling, where bigger samples are better). Which is exactly the case with Biedrins. We don’t know exactly what’s going on, and it’s probably the product of lots of things at work, including human psychology. What we do have is a much larger sample that tells us how good he is contradicting the small sample of the last couple seasons. The bigger sample has a lot more explanatory power, which is why it’s the best bet.

It’s every bit as likely he will relapse

That’s just it. No, it’s not every bit as likely. It’s less likely. Possible? Sure. Likely? No. Our bigger sample tells us that. To highlight this in example form: if you flip a coin 100 times, and you get 50 heads (so heads appears 50% likely), and then you flip the same coin 10 times, and get only 3 heads (so heads appears 30% likely), it’s much, much more likely that the coin will flip heads (in the future) more than 30% rather than less, and it’s more likely to be closer to 50% than 30%. 100 flips tell us more than 10. That’s exactly the point. This stuff really does hold true in reality. Biedrins much larger sample of FT’s from previous seasons tell us it’s much more likely he shoots better going forward than he has.

What specifficall are you saying?, "much , much higher" would be what? 25-35%? 35-45%?, 45-%5.5%? As far a number of total free throws are you projecting what? 3 times, 4 times, 5 times the current volume?

Just estimating, I expect him to shoot 50-55% going forward. With a computer and a lot of data, I could do a much better job of figuring that number out, but just quickly estimating that’s what I’ll go with. As for how many FT’s he’ll shoot? I’m not even going to try that. I do not know enough to even attempt to project it. Without an answer to how many he’ll shoot, I can’t give you an estimate as to what percentage he’ll finish at.