tag:blogger.com,1999:blog-12133335.post113364418220576080..comments2024-02-13T06:56:14.486-05:00Comments on Walk Like a Sabermetrician: Rate Stat Series, pt. 2phttp://www.blogger.com/profile/18057215403741682609noreply@blogger.comBlogger3125tag:blogger.com,1999:blog-12133335.post-1136079273684737982005-12-31T20:34:00.000-05:002005-12-31T20:34:00.000-05:00David, I remember your post on FanHome about this....David, I remember your post on FanHome about this. I was planning to talk about this in a later installment, whenever that will be. Thanks for reading.phttps://www.blogger.com/profile/18057215403741682609noreply@blogger.comtag:blogger.com,1999:blog-12133335.post-1136062128706986452005-12-31T15:48:00.000-05:002005-12-31T15:48:00.000-05:00So RC/O overvalues OBA, and RC/PA undervalues OBA....So RC/O overvalues OBA, and RC/PA undervalues OBA.<BR/><BR/>So why not combine them in some fashion? I remember that when I did this it turned out that equal weighting worked best. So we have outs plus PAs, which is (AB-H)+(AB+BB), or simply 2*AB+BB-H. To convert to "offensive games", find the empirical divisor to make the conversion, which turns out to be around 62.5.<BR/><BR/>So RC/G = RC/(2*AB+BB-H)/62.5Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-12133335.post-1135098238053886322005-12-20T12:03:00.000-05:002005-12-20T12:03:00.000-05:00What I was trying to convey is that the class 1 es...What I was trying to convey is that the class 1 estimators, like RC and BsR, are designed to estimate team runs, incorporating interactive effects. Therefore, if you apply these directly to a player, you<BR/>are asking the question "how many runs would he score if he was a team". And for teams, R/O is clearly the correct rate stat.<BR/><BR/>In the class 2 and 3, you are trying to measure the change this guy has on his team's runs scored, and therefore your rate stat should reflect how he affects team runs scored. R/O is answering the team question.<BR/><BR/>Of course, the distortion involved here is pretty small. I use R/O on my website to find RAA or RAR, and so does just about everybody else, and I don't think it's really a problem, especially for real players. Even the Bondes,<BR/>Ruths, and Williamses, as far as I know, don't really cause any problems. But in the interest of a rate that is fully applicable to any possible hitter, I don't think R/O is technically correct with LW or TT RC, etc. But using it does not really screw things up, as far as I can tell. The rest "rate stat<BR/>series" is pretty low on practical value.phttps://www.blogger.com/profile/18057215403741682609noreply@blogger.com