John Hollinger from ESPN, who originally gave the Denver Nuggets and Golden State Warriors almost a zero percent chance of advancing past the first round based on his prediction model, admits now that his data might have been a little skewed.
Hollinger, who makes his career by doing bad statistics with bad assumptions (which in turn creates bad models), always acknowledges that his models are based on assumptions, and things such as injuries, streaks and "tanking" may not correctly reflect the true odds. However, his correction is still not accurate according to his logic. He throws out all data points for the Nuggets and Warriors if they had a significant player injured, if they didn't need to win or if the team they were playing didn't need to win or may have been trying to lose. But he didn't do this for their playoff opponents, which means the Nuggets/Warriors records get inflated because of bad situations and the Spurs/Mavs don't get this same benefit. Although those numbers would greatly improve Denver and Golden State and only mildly improve (if at all) Dallas and San Antonio, which means his corrected numbers are a lot better than his old ones. The point is, his logic is always a little misguided.
My main problem with Hollinger is his famous PER, which supposedly is the tell-all of how good a player is based on efficiency. Of course he doesn't publish his formula, except in a his book that he expects me to buy. Having said that, I don't know what he emphasizes in his PER, and I haven't really tried to figure it out, but I do know that there are several things wrong with it. The most glaring error is Steve Nash, who's having his best season and one of the most efficient seasons I've ever seen as an NBA fan. Steve Nash is apparently the 12th most efficient player in the NBA according to Hollinger's numbers. If you base this on offensive efficiency, Nash should easily be number one. His fg% is .532, which of the 11 people in front of him, only Duncan, Gasol and Boozer are better, and these are guys who spend a lot of time under the basket getting easier shots. Not to mention that in true shooting % (which factors in that 3-pointers are worth more) Nash is at .654, which is a lot better than anyone in front of him. In fact, he's number 2 overall in that category behind Brent Berry, who basically only shoots wide open three pointers and doesn't exactly have any pressure to create offense. So basically Nash is the most efficient shooter in the NBA, and as far as other offensive stats go, there would be assists (which Nash dominates) and turnovers. The turnovers for Nash can be easily factored into an assist to turnover ratio, which is 3.08 for him, much better than anyone ahead of him on the list. The only major category which Nash isn't relevant is rebounding.
I don't know how Hollinger got his numbers, but it just seems to me that the best shooting player in the league who also leads the league in assists and who might be in the top 5 all-time efficient players, should probably be higher than number 12 in the 2006-07 PER.
No comments:
Post a Comment