December 22, 2015
We Can Be Heroes?
I was recently listening to some back episodes of the MLB Statcast podcast (love it!) and heard a fantastic interview between host Mike Petriello and former major-league outfielder turned Rockies broadcaster Ryan Spilborghs. (It’s the 11/5/2015 episode, which you can find here). In the interview, Spilborghs talks about how a team’s approach might make all the difference between winning and losing. In particular, he talked about how, when a team is going bad, guys get a little desperate. Instead of taking a nice rational approach at the plate, they go up attempting to hit a home run to try to light a spark under the team.
It’s an interesting hypothesis. Players are human, and most certainly they walk from the on-deck circle to the batter’s box with both a bat and their own emotional baggage and their memories of last night’s disappointing ride back to the hotel. And if it seems that everyone else on the team is scuffling, there’s probably at least a temptation to try to do something to make things better. Maybe if Statcast had mind-reading capabilities (just a suggestion, guys… just a suggestion), we’d see that players really do go up with a different mindset when things haven’t been going well. And maybe they are thinking “Fiddlesticks! I’m just going to try to hit the ball 500 feet.”
But does it make a difference?
Warning! Gory Mathematical Details Ahead!
For a moment, let’s take a detour into a seemingly unrelated question. What happens to hitters in extra innings? The link between the question of a team who’s been on a bad run of form and a team playing in extras is that there’s a good amount of hero potential in both. In extra innings, a home run could win the game right then and there. Even in the top of the inning, a home run is going to go a long way toward ensuring a victory. There’s plenty of incentive for a hitter to walk up to the plate with the goal of hitting a home run. It’s also reasonable to think that if he did go up with this idea in his head, he’d actually do something that actually made it more likely for him to hit a dinger, even if the other effects it produced were net-negative to his overall chances of doing something helpful for the team. So, we should at least see that in extra innings, home runs should increase.
I looked at all games from 2011-2015, and using the log-odds ratio method, created a baseline expectation, given a particular batter-pitcher matchup, that a plate appearance was to end in one of several outcomes (including home runs, strikeouts, walks, and a few others). I then entered this control variable (in log-odds form) into a binary logistic regression. I added a dummy variable for whether the plate appearance took place in extra innings or not. The control variable is bound to be extremely significant (it’s designed to be), but the dummy variable tells us something interesting. If more (or fewer) home runs are massed into extra innings, the dummy variable will let us know that.
It turns out that that there’s a slight downward effect on home run hitting in extra innings. The effect was not statistically significant (p = .273), but the trend line pointed downward. Same story, complete with non-significant findings, for extra basehits. (This might be because teams are playing “no doubles” defense.) However, there is a significant increase in singles in extra innings, compared to what we otherwise might expect. The same can be said for flyballs (they become more rare) and groundballs (they become more common). Seems that while we might predict that the context would make players more flyball happy, hitters are actually more likely to become groundball/singles hitters in extra innings. Players are actually trying for the lower reward, but higher probability of a positive outcome approach!
So, let’s come to the question of whether hitters try to play the hero when the team hasn’t been doing so well. Using the same basic method as above, I created control variables for various outcomes and then calculated the team’s winning percentage over the previous 10 games that it played. I entered that into the regression to see whether that made much difference.
Did players hit more home runs than we would expect when their team had lost seven of the last 10? It turns out that the only significant effects were in strikeouts and walks. The more successful a team had been over its past 10 games, the more likely its hitters were to strike out in game 11, and the less likely they were to walk. When things had been going well, hitters actually got sloppier (and conversely, when things had been going badly, they actually seemed to respond by being more patient and working walks!)
It’s an interesting finding that when things are getting tense, players might feel frustrated with their teammates or they might feel a desire to play the hero. But their actual behavior suggests that they become somewhat more conservative (tentative?) in their approach when the chips are down.
Why So Tentative?
It’s worth noting here that these analyses use the large-N “all players” model that treats all players as a big unindividuated mass. It’s entirely possible that some specific players react to adversity by wanting to play the hero and some by becoming more tentative. Maybe those two groups cancel each other out. But in the aggregate, “players” actually don’t respond to adversity by pushing on the gas pedal harder. They actually become a little more circumspect and in some cases, choose less risky strategies with lower rewards.
Why would this be, when standard alpha-male psychology would suggest that faced with a challenge, the response should be a Home Improvement style, “More power!” The answer might lie in what’s known as a rebound effect. The first reaction might be “I’ll show everyone! I’m going to hit this ball 500 feet!” Then, there’s another little voice inside that says “Y’know, as much as I would love to hit this ball 500 feet, the chances that I’d actually be able to do that are actually kinda slim. I’d do more for my team by getting on base and giving up the macho garbage.” In other words, the voice of reason takes over. And a hitter may even be overcompensating for the first thought. If a hanging curveball presents itself, you want to be ready to hit it into the nearest body of water, but it’s not likely that a hanging curveball is actually coming down the pike anytime soon. They just don’t happen very often at the major-league level.
But it’s clear that the idea that players go into “hero mode” when faced with adversity isn’t actually supported by the evidence.