Effect of SB attempt on batter (November 10, 2003)
From those who don't remember, Bill James published a study from a reader of his. It's been excerpted here
(Do a search for Driscol)
John Jarvis took a look at it a few years ago.
As did Mark Pankin (PDF file)
--posted by TangoTiger at 11:47 AM EDT
Posted 1:29 p.m.,
November 10, 2003
(#1) -
RossCW
None of the studies seem to consider the quality of the pitcher and its impact on when batters choose to steal. Without that any evaluation of the impact of stealing on the batter may just be catching variations in the quality of pitching. Henderson attempts steals against all pitchers, Murphy only when this team is playing for a single run against a quality pitcher.
It seems to me one of the fatal flaws in a lot of sabermetric analysis is failing to account for the fact that players/managers are consciously optimizing their chances of success. You need to be very careful that the effect you are measuring isn't just an artifact of that optimization.
Posted 2:41 p.m.,
November 10, 2003
(#2) -
JHP
I remember STATS Scoreboard doing a piece on something similar, in I believe the 1994 Abstract. They studied the effect of pickoff throws on the HITTER and concluded that it's the batter who gets distracted and rattled by a dancing baserunner, NOT the pitcher.
There's definitely plenty to look into here. How many of the SB attempts were hit-and-runs? A botched hit-and-run would likely account for at least some of the outs.
As for the optimization, that's the point. The idea is to find out the numbers that actually explain what the chances for success are. The "old-school" methods come under fire precisely because they don't increase the chance of success.
I refuse to believe that the stolen base is useless, but if utilized properly, there's nothing wrong with taking an extra base, as long as you don't run yourself out of an inning/rally. The Bule Jays data is fascinating though. Who was stealing and who was at the plate when you crunched most of the numbers, i.e. did Delgado hit .171 after a SB too?
And Jarvis's piece was a little too numero-intensive for me to take a look at here and now...
Posted 2:54 p.m.,
November 10, 2003
(#3) -
Jhat
(homepage)
I didn't collect the numbers, that was done by a gentleman named David Driscoll in 1985.
I asked Bill if this had been followed up: he said that it probably should have been, but the stolen base was dying at the time so it didn't happen.
Posted 3:19 p.m.,
November 10, 2003
(#4) -
Ted T
What's definitely true in the data is that the change in batter performance with runner on first only versus bases empty is largest (by a significant amount) in aggregate for batters in the #2 slot, followed by the #3 slot. For slots #4 through #1, the change is not significantly different from constant (but positive).
My pet interpretation of this data is that we must be seeing some difference in defensive conduct here related to the identity of the runners on first in these situations.
Note that of course runner on first only is basically "conditional on not attempting a SB". As is well-known, conditional on a SB attempt, performance does decrease.