For baseball it is a good indicator on over/under performance. Normally teams will return to the mean over a large sample size. The Rockets would be playing right at the expected level based on +-
Gotcha. That's a good point. So if they were way over their expected W-L, there'd be a concern that they'll return to the mean. Alternatively, I recall some teams where they throughout the season under/overperform. The Rubio/Love TWolves (especially one of the squads 4 or 5 years ago) in particular... where +/-, etc. all said they should be better, but they never reverted to the mean, cause they always sucked in the 4th. I guess OKC would be like that this year so far, and the question is when do they revert to the mean and stop losing so much?
Pythagorean win shares work better in sports like Hockey and Baseball, due to less individual impact and fewer chances to score. What it is very good for is showing the difference between contenders and pretenders with similar win-loss records. The inverse applies to bad teams like those T-wolves.
The man is brilliant, and not just because he supports Batman's theatre in Houston. He doesn't hit it out of the park from time to time, but I'm damned glad we have the guy as our GM.
just for fun based on Morey's formula projected season wins as of now GSW 80% 66 wins HOU 79% 64 wins BOS 75% 61 wins OKC 63% 52 wins SPURS 60% 49 wins CLE 52% 43 wins
I still want Morey to be fired even this team is wining so far, because this is not a championship team, situation similar to Boston.
I'm not a stat student. So just guessing, I suspect they looked back at tons of historical teams and seasons, and this is the formula that had the highest correlation. But maybe not. And even if so, it is curious why!
I just figured out the current prediction and it's accurate even now. The plus minus with the 14th power thing comes out to a .791 record so far. Multiply that by 20 games and you have 15.82 victories. It's voodoo if you ask me! Works on Boston too! They're predicted to have 18.25 wins right now.
There isn't any deep meaning to the exponent itself, and I'm not sure if Morey came up with "14" or he had worked out a version in his early career that had some other exponent. I believe others (like Dean Oliver) came up with a similar formula in their work independently of what Morey did, so I don't think it is something he can take exclusive credit for (and I don't believe he has, although that picture and the article it comes from makes it seem otherwise). It can be derived through a regression analysis over a ton of NBA seasons. Basketball-reference came up with 14 by doing this over all BAA, NBA, and ABA seasons: W Pyth Pythagorean Wins; the formula is G * (Tm PTS14 / (Tm PTS14 + Opp PTS14)). The formula was obtained by fitting a logistic regression model with log(Tm PTS / Opp PTS) as the explanatory variable. Using this formula for all BAA, NBA, and ABA seasons, the root mean-square error (rmse) is 3.14 wins. Using an exponent of 16.5 (a common choice), the rmse is 3.48 wins. (Note: An exponent of 10 is used for the WNBA.)
