John Jarvis SABR presentation on the IBB (August 1, 2003)
I will make my comments after reading this...
--posted by TangoTiger at 01:01 PM EDT
Posted 1:45 p.m.,
August 1, 2003
(#1) -
OCF
This looks at first glance to be excellent work.
There is one element I'd like to see added to the analysis. Everything here seems to be geared to analyzing runs scored in the inning in which the IBB was (or wasn't) issued. What I would like to see is an extension of the analysis to the following inning as well. In every case, the IBB "rolls the lineup forward" and thus potentially creates more plate appearances for good hitters. But there's one case that makes the need to look at the next inning acute, and that's the IBB to the #8 hitter in the NL. One of the charts in this study says that this is an extremely common IBB situation, and in particular is a large fraction of the lowest-SLG batters being walked. Since pitchers are, on average, very poor hitters, this has a high probability of "working" as far as preventing runs in the inning of the walk. But there has to be a large difference in run expectation for an inning with the pitcher leading off and an inning with the leadoff hitter leading off.
Posted 2:00 p.m.,
August 1, 2003
(#2) -
tangotiger
Excellent point on the "rolling-forward" concept.
I have my own method of handling the IBB issue, and it has similar variables to Jarvis'. I do not use SLG or BA, but rather a tailored version based on the 24 base/out state that we are in (much less when looking at the IBB situation). I would definitely use the LWTS by 24-base out states, rather than SLG and BA.
I'm also disturbed that Jarvis only looks at the recent performance to establish the player's performance level, because the regression deemed that to be the best fit. What you should use is the best estimate to a player's true talent level. Maybe at a high group level, the 2-week performance does the job, but at an individual player's level, it definitely can't.
All in all, I'd have to give this thing a few more readings.
Posted 8:59 p.m.,
August 4, 2003
(#3) -
David Smyth
I'm surprised that this thread has not gotten more of a response. Jarvis is, as far as I can tell, the leading analyst on IBB. He has more stuff on it than anyone else. I think that there is a ton of information and possible insight in this article...