Brett Green is an assistant professor at the Haas Business School who, in tandem with Stanford University’s Jeff Zweibel, recently wrote a paper in the controversial field of “hot hand” literature. The “hot hand” is a concept in sports that states a player can become “hot” and temporarily increase his level of performance.
The argument entered the mainstream in the 1980s, when a paper written by Thomas Gilovich and Amos Tversky posited that the hot-hand was a cognitive fallacy. Green’s paper — which concluded that evidence of “hotness” or “streakiness” does exist — contradicted the Gilovich and Tversky conclusion. There are real, tangible effects in the sports themselves. Depending on a coaches’ view on the hot-hand, it affects who gets to play at the most crucial points in the game. Say the coach pulls a starter for a moment in the fourth quarter and a bench player hits three shots in a row. Do you leave the bench player because he has the “hot-hand,” or do you put the starter back in because he has a higher baseline of talent? The issue divides traditionalists, who swear by the hot hand, and progressives, who will go to their graves denying its existence.
Daily Cal: For those who don’t know, could you explain the concept of the “hot hand”?
BG: People sort of have different notions of what the hot hand is. What we do in our paper and what most people think about when they think about the hot hand is that there’s some time variation in ability over the course of a season. Some of the time, you’re in your normal state, and, say, your shooting percentage is 50 percent. Every once in a while, you have the “hot hand,” where your shooting percentage goes up to 55 percent. Other times, you have the cold hand, where your shooting percentage goes down to 45 percent. Over the course of the season, you might oscillate between these different states.
DC: And what is the goal of your paper?
BG: The methodology is essentially to test the hypothesis that there is no hot hand. Our null hypothesis (the basic assumption of the study before research is conducted) is that there’s no hot hand, that there’s no time variation in ability and ability is constant over the course of the season. What we do is reject that null hypothesis based on the patterns we do see in the data. There’s enough in the data that tells us with very high confidence that we can reject this null hypothesis of no hot hand.
DC: It’s been out there for a little bit of time, right?
BG: I’ve had the results since 2006, but it took us a while to write it. You get busy with other stuff, and some stuff goes on the side burner. It’s a side project, to some extent.
There’s not a lot of economists thinking about these things. I think there’s some scope for academic research in sports economics. I mean, there are people doing these things, but there’s not so much being done that there’s no papers left to write.
DC: Do you have a guess as to why the hot hand exists?
BG: One reason that people don’t associate with the hot hand is physical health and well-being. Baseball players, if they’re healthy, play 162 games. Some of the time, they may be injured, and they play through injuries.
I think part of it is physical and part of it is mental, but that’s from my introspection of playing sports — baseball, tennis, golf — and having these feelings of confidence, whether I was standing over the ball or facing my opponent. So I believe there is a mental aspect to it. I don’t know why.
DC: It seems, based on what you’ve found, the criticism of managers swearing by the hot-hand has been off-base.
BG: I think (Cleveland Indians Manager) Terry Francona has a much better sense of the underlying model than I do. His decision-making is way better than me running all these models.
My underlying view of the world is that good baseball managers are probably getting it right. If they weren’t getting it right, they wouldn’t be managers for very long.
The fact that many managers are doing something along these lines is a very powerful phenomenon that, from my perspective, would be hard to deny. It’s surprising that we haven’t found any evidence, given what managers, players and fans all think.
DC: Do you think it’s possible to be able to prove that the mental feeling of the “hot hand” exists?
BG: It’s not easy, but it’s not completely hopeless. There are strategies that will allow you to get at this question, especially given the explosion of neuroeconomics.
BG: We’re starting to run brain scans and trying to figure out what exact parts of the brain are firing to predict certain types of behavior. It’s still a little bit too costly to do frequently, but in 20 to 30 years, we’re going to be using this to do all sorts of things that we’re not doing right now. That’s one way you can try to get at this, but there are other ways that require a bit more creativeness but are potentially out there.
DC: Have you received any hate mail from anti-hot-hand truthers?
BG: I wouldn’t call it hate mail, but we did have some pretty heated back and forths with some people in the community of sabermetrics (statistically-minded analysis of baseball).
DC: Which ones?
BG: There are two sabermetricians — Tom Tango and Mitchel Lichtman — who are co-authors of “The Book.” Lichtman ran a study on streakiness and found that hitters aren’t streaky. When our paper came out, there was a lot of dialogue on (TangoTiger.com). And this was great; it critiqued what we had done, and we got a lot of valuable feedback from them.
The first question we wanted to understand was why we were getting different results from this book. We think we were able to come up with an explanation for this difference, which we explained to (Tango) and (Litchman). And to an extent, I think they agreed with the explanation. However, we weren’t able to get them on board with our arguments that there is evidence of streakiness. They were calling it something different — they were calling it “recency.” But when we finally got down to it, they agreed that “of course there is this pattern in the data.”
DC: Do you consider yourself sabermetrically inclined?
BG: Not really. I mean, it depends on your definition of sabermetrics. But based on the back and forth we’ve had, I’d say that sabermetrics and economics are pretty far apart.
DC: Why do you think that?
In many cases, the objectives of the two fields are different. Sabermetricians are primarily interested in predicting outcomes and finding metrics that help them do so. We (as economists) often care more about the mechanism underlying the prediction. To do that, we write down models and form hypotheses about how we think the world works and then use data to try to test those hypotheses.
DC: And by models, what do you mean?
BG: I mean, what is streakiness? Write down a mathematical model of streakiness for me. This is the first thing that we need to do before trying to answer the question of whether streakiness exists. Surely the model will not be an exact description of reality, but it is useful for guiding what tests we will run on the data. We thought, “What does streakiness look like?,” and we formed a definition and then wrote down a simple model of it. Without an underlying model or set of models, the interpretations of the data end up being a lot more subjective.
DC: Were you surprised at the results you found?
BG: Not really. As an athlete, I had a belief that this effect is there. It’s more surprising to me than not that it hasn’t been identified.
I think there is some overestimation of the effect. We’re talking about a 1 percent increase, which is about the difference between a hot and an average batter, so it’s not massive, but it is when you’re talking about baseball, when a .270 hitter is average and a .285 hitter is well above average.
It’s important for decision-making on the margin. You wouldn’t want to ever sit your all-stars, but you might want to sit someone who’s on the margin — between starting and not — if he’s not hitting and play him when he’s done well recently.
And I think this is consistent with how managers operate. I don’t think they sit David Ortiz very frequently because of poor recent performance.