There will be a very short planned maintenance outage of the site tonight (7/22) at 11 PM ET
April 22, 2014
The Houdini Hangover Effect
Rainbow sprinkles alert: Ben Lindbergh saw this one on Twitter, from currently shelved reliever Peter Moylan, who was traded to the Dodgers in the middle of last year after spending several years with the Braves. Mr. Moylan is currently recovering from Tommy John surgery, like everyone else in baseball.
Well, is it?
Warning! Gory Mathematical Details Ahead!
I controlled for the strength of the reliever and the opposing batter using the log-odds ratio method to calculate the expected probability that a plate appearance would end in one of several events (strikeout, walk, single, etc.). I also controlled for the reliever’s pitch count (maybe the reason that he got out of the jam was that he got himself into the jam to begin with) and whether or not he had a platoon advantage (a good reason that he might come out for the eighth is that he might be used as a LOOGY for one more batter). I ran a series of logistic regressions, looking again the same basic subset of plate appearances, a reliever pitching late in a close game (again, pitching team either winning by less than three, tied, or down by one) so that we’re comparing similar sets of contexts. I entered whether our reliever was a “returning Houdini” into the regression to see what effect it might have.
The results of those regressions were loud and clear in their message. For returning escape artists, strikeouts were well below expectations, as were outs in general. Walks, singles, and extra-base hits all increased, although home runs decreased (not enough to offset the other effects). The effect sizes were huge. As an example, take a situation in which we might otherwise expect a 20 percent chance of a strikeout. For a returning Houdini, the chances of actually striking the hitter out fell to around 12 percent. Other effect sizes were similar. Managers take note?
Not so fast. Consider the fact that one reason that a pitcher might have gotten himself to where he needed to escape was that he wasn’t really on his game that day. I further restricted our Houdini sample to those who faced (and conquered) a tough situation, but who had inherited the runners on base, meaning that while they escaped the mess, they had not been the ones to create it. Once I did that—again controlling for batter and pitcher quality overall, pitch count, and handedness advantage—the effects completely disappeared. It seems as though the reason that our Houdini pitcher shouldn’t bother coming out for the next inning is often that he just wasn’t having a good day to begin with.
Escape is Never the Safest Path
But once our pitcher has escaped, has run back to the dugout, and is sitting there with a jacket on to keep his arm warm, his brain recognizes that he is out of “danger.” The body activates the parasympathetic nervous system and its associated messenger chemical acetylcholine, which restores a resting balance. It takes a few minutes, but he probably has 10-15 minutes to sit there, and by the time he’s able to pitch again, he’s probably returned most or all of the way to a resting state. The reason that the body has such a system is that, physiologically, it can’t sustain the fight-or-flight changes for extended period of time without risking other negative consequences. That fading of the adrenalin rush might feel like a letdown, but it actually helps our Houdini reliever to go back out there for the eighth and not burn out.
In fact, the numbers show that our reliever is just as effective as we might expect him to be, given what else we know about him and his current situation. At least the average reliever is—there probably are differences between players as far as how much a pressure situation takes out of a pitcher, but I would argue that the burden of proof would be on the side that says that Smith absolutely needs to come out after an escape act.
The more important thing is that an escape act, while nice, seems more likely to be the sign of a couple of lucky breaks for a guy who was having a bad game. If a guy who pulled an escape act created his own mess, he probably should be replaced, or used with extreme caution. But if he was just a fireman who cleaned up someone else’s mess, he’s good to go if you need him.