by Kevin Harlow
April 18, 2004
Derek Zumsteg's April 13, 2004 Breaking Balls made me ask myself the question "What do you do with BA/HR/RBI when you're given it at the ballpark?". OK, so that's a philosophical baseball question that may include Homer Simpson as part of one of the options. Another option would be to convert it to OPS. I'm not saying that it's a good option, but here's how you might go about doing it.
First, look really hard at the RBI. When you start seeing spots, close your eyes and tilt your head backwards, then you will see ... sorry, I got my options confused. Seriously, just glance at the RBI column long enough to know where it is so that you can avoid it.
OPS from Avg and HR (and G)
1) Guesstimate the player's HR pace over 81 games. Yeah, I know, this is the hard step. Now divide it by 100.
2) Double the batting average.
3) Add together #1 and #2 and 0.13.
Example:
For example, on April 18, 2004, Ken Griffey, Jr.'s line is .250/3/7 in 9 games, with an .873 OPS
1) 9 games into 81 games goes 9 times. 9 times his 3 HR is 21 HR. 21/100 = 0.21.
2) Doubling his .250 average gives .500.
3) OPS ~= .27 + .5 + .13 = 0.900, which puts you in the ballpark, so to speak.
How accurate is this? If you could perfectly estimate HR/81G, this method yields an R^2~=0.9 using all individual batting seasons from 1962-2003 with a minimum of 400 PA. What, you don't think you can estimate it exactly? It's not a problem. An unbiased HR/81G estimate with a normally distributed standard deviation of 5 HR/81G still yields an R^2=0.84. Given that just using Avg, as in something like OPS~=3*Avg-0.05, has an R^2~=0.5, any rough estimate of HR/81G is going to significantly improve the accuracy.
Go to Kevin Harlow's home page.