Re: OPS vs. OBP vs. BAS
This thread got me curious, so I ran a quick and dirty regression on a few key statistics and how they correlate with runs scored per team. I simply put in runs scored as the dependent variable and made the other stats the independent variable. The best measure of correlation (R-Squared) tells us how much of the variation in runs scored is explained by this variable. Below are the results (and the P-values are very low for all items below):
Batting Average - .64 (that's not very good and this model, in most cases, would be considered very poor in it's predictive value). Also, the residual plot showed significant autocorrelation.
OBP - .76 - Still not great, but better than BA. Residuals look OK on this one.
SLG - .86 - Getting warmer
OPS - .94 - Folks, we have a winner
BA w/ RISP - .56 (basically, no predictive value)
Strike Outs - .003 ( :MandJ: )
Walks - .36
And the standard error for OPS is only 40% the size of the SE for BA.
Also, if you compare a team's BA with RISP and compare it with the team's overall BA, there's almost NO difference across the board. Of the 30 teams, only 4 had a difference greater than 5% from their overall batting average. Only 1 team had a difference greater than 10% (which happened to be 10.9% and it was Milwaukee and it was lower).
Grape works as a soda. Sort of as a gum. I wonder why it doesn't work as a pie. Grape pie? There's no grape pie. - Larry David
Last edited by MWM; 12-13-2004 at 01:01 AM.