Bucks Diary: Is 82games.com statistical nonsense?

According to 82games.com Zaza rules

Ever since an article appeared in the New York Times last spring touting the new statistical revolution in the National Basketball Association, the website 82games.com (mentioned prominently therein) has been the buzz of hoopsheads everywhere. The question I have is, do the statistical conclusions contained on that site have any value?

Let me say first that I am not one of these guys that instinctively attacks number crunching sports analysts nor do I reject the work they produce out-of-hand, and I certainly try to refrain from labeling them pocket protecting geeks. I genuinely think good solid statistical analysis can give a competitor the upperhand in almost every industry, so I welcome it. In this case, however, I'm concerned that what 82games is pushing is not good analysis.

I don't pretend to understand the convoluted formulas they use to rate players and their overall contributions to their respective teams. Its so user unfriendly you would almost have to take a full semester course to master it. (By the by, I've found that's always a bad sign when it comes to statistical credibility). But from what I bothered to learn about it, I gather the numbers their formula generates reflect the contribution a given player makes to his team according to how that team does when the player is on the court versus when the player is off the court.

Well first of all, I teach a logics class so logical flaws sometimes seem obvious to me. One jumps out at me here. It is what I call the "coincidental occurence" and the "happenstance" fallacies. That is, simply because two things occur at the same time does not mean that one caused the other, and simply because someone produced certain results doesn't necessarily mean you can extrapolate those results and consider them fair evidence of that person's value. Yet that is sort of the baseline logic behind 82games statistical analysis.

Let me give you an example. You find you need to undergo dangerous surgery. The hospital recommends a certain doctor perform the surgery. You look at the hospital's statistics and find that the mortality rate for this doctor is considerably higher than the overall norm for the staff. You are understanably concerned that this doctor is not skilled. That's essentially what 82games would have you believe. Yet it turns out the doctor is actually the best surgeon on the staff. He is so skilled that he is the only one on the staff who can perform the most risky cutting edge surgery, accounting for his high mortality rate.

Like the surgeon example above, 82games is saying because the team as a whole performed in a certain manner while a certain player was on the court, and in a certain manner while that player was off the court, those differing performances can be reflected back upon the player. Over time perhaps that may be true, but you must be very, very skeptical. There are so many variables inherently involved between the cause-and-effect they want you to buy, I don't know if its a good purchase. When I look at analysis and judge its merit, I want to know the specific reasons the conclusions occured, what 82games gives us is mere happenstance.

If their logic is faulty, some of their conclusions are shocking. For instance their numbers from last year indicate the Bucks were worse off with Michael Redd on the court. I find that hard to swallow. Then if you look at their numbers from 2003-2004, you would be led to conclude that the Bucks were much better off with ex-Buck Erick Strickland running the point and burgeoning superstar point guard TJ Ford on the bench. Come on now.

If that weren't enough to warn you off 82games' brand of basketball analysis, then get this: if you put faith in their numbers it would seem former Bucks center Zaza Pachulia is on his way to making the next 50 Greatest List. That's to much. Anyone who's watched him play over any stretch of time knows that marshmallow has no game. Yet somehow 82games numbers consistently indicate that he is a very valuable commodity wherever he goes. That's a joke. I checked the boxscore from last night's Hawks game and superstar Pachulia -- a center remember -- somehow went 1-for-9 from the field. Was the basket moving around Zaza? Do you see why I might be leery?

Maybe 82games.com has some value. It probably does. But I haven't found it, and if I were a GM in the NBA, I would certainly view their information with a grain of salt.