This is a thought per two articles I've read recently.  The first was Dave Cameron on Fangraphs asking where Sabermetrics should go next.  The other was Larry Stone's interview of Don Wakamatsu in the Maple Street Annual.

Specifically Wakamatsu said that part of why he bunts is, "everyone on the bench has to get off their seat when a guy [bunts].  That's the dynamic that's not in the numbers."  My initial instinct was to dismiss this sort of thing as silliness.  Whether or not you jumped out of your seat in the previous inning should have no bearing on your performance going forward.  But then later, Wakamatsu said when referring to player platoons, that a sense of fairness comes into play.  And this made me think of managerial actions through the lens of behavioral economics.

Specifically I thought of the dictator game which essentially looks at how people respond to being given a portion of a fixed allocation of resources.  In the game there is a fixed amount of resources (say, 100 dollars) and 2 players (players A and B).  Player A allocates themselves a portion of the resources leaving the remainder for Player B.  Player B then gets to accept the allocation given to them, or reject it and both players wind up with nothing.  In a perfectly rational world Player B would accept any non-zero number; they would be getting more than the nothing they had initially.  However, in experimental settings, the threshold for rejection was far higher than zero (usually in the 25-30% range).  Basically what this shows is that we’re biased against decisions we perceive as unfair, and this bias will lead us to act in ways that are detrimental to our own performance.

Now this game doesn’t correlate perfectly to divvying up at bats in a platoon setting, specifically the idea of rejection and the consequences of rejection are quite different, but I think that one could model a game such as this to investigate the psychological aspects of baseball more deeply.  How would you measure managerial trust?  I’m not sure.  But I bet you could come closer to quantifying its value.

Part of the appeal of baseball to the statistically inclined part of my brain is that it is a deeply controlled game.  Each interaction is both able to be quantified and discrete (pitcher/hitter matchup, ball in play, baserunning decisions).  Also, the end goal is very simple: win.  How do you win?  Score more runs than your opponent.  Runs and run prevention are perfect measures of utility.  Every action on the field has a definitive measure of utility.

But with all of that said, baseball players, like all people are irrational.  Perhaps more so.  And this is where Dave’s article comes into play.  I think there are two related questions that haven’t really been sufficiently addressed in a scientific way: how do you motivate players to behave in a utility maximizing fashion (get them to want to bunt only when bunting makes sense statistically), and if some of those techniques involve in-game strategies that do not maximize expected run value, are there times when those seemingly irrational strategies are actually preferable?  Maybe the motivational force of a borderline bunting decision actually has sufficient value in player performance to serve as a rational decision in an irrational world.  I don't know, but I bet we could find these answers.

Continue reading this post »