Considering High Leverage Performance and Clutch Hitting by Neil Weinberg December 5, 2014 Human beings love big moments. We have an innate attraction to crescendos, buzzer beaters, walk-offs, and those scenes in movies when people sprint through airport terminals. It matters to us in a very primal way what transpires when the chips are down. This is why RBI is a popular statistic and why so much attention is paid to stats like batting average with runners in scoring position. We believe that players who perform well in the big moments are the best players. There are probably all kinds of cognitive and psychological biases at play, but I think we can all agree that success in critical situations is more highly valued than success in general. This is as true in life as it is in baseball. Yet there is also a lot of evidence that tells us to ignore these performances in baseball, or rather, to treat them just like any other performance. A home run with the game on the line is more important than one in a blow out, but it’s not really a reflection of the player being better or being clutch. This is a controversial stance. Sabermetricians have been commenting on the false “clutch” narrative for many years and have received a great deal of push back. The alternative view is that certain players are able to rise to the occasion and that they know how to slow the game down and deliver in critical spots. Rather than taking a hard line on the subject rhetorically, instead I’d like to review a bit of the research done on clutch and provide some important questions to consider regarding clutch performance. A couple things are important up front. First, clutch hits exist. Some moments are more important than others and getting hits at those times is very valuable. The contention is not that hitting in the clutch isn’t valuable, the contention is that it is very difficult to look at a player’s performance and learn very much about his true ability in the clutch. This isn’t to say that individual hitters don’t change their approach in clutch situations, as Russell Carleton noted, but rather that it’s just really hard to find evidence that certain hitters consistently do better in clutch situations. Some hitters might do better, but it’s rare for a player to even receive 100 high leverage PA in a season, so the odds that we can glean much from his performance is quite low. This piece by Joe Sheehan does a nice job summing up some of the foundational research, citing studies that show there’s virtually no evidence that players who are clutch one year will continue to be clutch in the future. Here’s an older study from Harold Brooks that comes to a similar conclusion. An old post from Tom Tango comes to an instructive conclusion: There were two observable clutch hitters from 1999-2002. Here’s a good discussion from Phil Birnbaum about the proper way to think about clutch, as a very small effect if anything. You may be familiar with the small clutch effect found in The Book, but Birnbaum also offers reasons to be skeptical there as well. Tom Tango offers a more recent account, saying that yes clutch hitting exists, but that the effect is probably pretty small. These aren’t the only articles written about clutch, but they link to many others and you can find some others on your own as well. They offer you a sampling of what’s out there and leave us with a pretty clear set of results. Clutch hitting, at most, has a small effect. There might be a couple of players with some type of preternatural ability to perform in tight spots above the level of the average pro athlete. And there might be players, who if given a larger enough sample, do consistently hit a little better in clutch situations. The exact method determines whether you arrive at statistically significant results, but no one has found a big effect. Clutch hitting isn’t necessarily nonsense, but it’s also not the kind of thing you can discover by looking at a season’s worth of data. Perhaps certain players are capable of excelling in important spots, but it takes a long time for us to be confident of that fact. Now that we’ve looked at the research to date, I’d like to simplify the conversation and talk about a couple of key factors that you should consider when deciding how much you believe in clutch hitting. 1. Players Don’t Get Many Clutch Chances Each Year It’s entirely possible that clutch hitting exists and that some players can rise to the occasion when others can’t. That’s a logical possibility and an empirically testable question. But the problem is that most hitters get 60-80 high leverage plate appearances a year. A few get close to 100. We know from lots of other areas of research that a sample size of PA that small just can’t tell you much about performance. Maybe strikeout or walk rate. Maybe something about plate discipline. But 80 PA just isn’t enough data to tell you about something like wOBA or wRC+. As a result, we need to make this alteration to the clutch doctrine: Clutch hitting may exist, but a single season of clutch hitting does not provide evidence one way or the other. 2. We Can’t Really Define Clutch Let’s again say that clutch hitting might exist. The problem is that we don’t know which situations the batter perceives to be clutch and batters also likely do not think in terms of bins that say “not-clutch PA” and “clutch PA.” Pressure is a continuum and players probably have different ideas about big moments. We define clutch with Leverage Index in most cases, but it could be based on the number of runners on base exclusively. Or it could be conditioned by who is on the mound or any other collection of variables. To determine if clutch hitting is real, we need an independent variable that properly measures clutch and we have no way of knowing if our definition is accurate or if we have properly categorized the situations. Should we distinguish between an LI of 1.9 and 2.4? Do batters? As a result, we need to recognize that our findings are only as good as our inputs. Even with a big sample size, defining clutch is a serious problem that doesn’t get enough attention. 3. Clutch Hitters Might Be Lazy If you’re an intelligent person, and I bet that you are, you know that you shouldn’t take the absence of evidence as the final word. Just because I can’t find clutch in the data doesn’t mean that it isn’t there. All that means is that the evidence currently supports its absence. New evidence could change that. But now we need to consider clutch as an idea. The idea of clutch hitting is that some hitters perform better with the game on the line. But what does that say about those hitters in other situations? Are they not trying very hard? Do they only focus when the flashbulbs are going off? If you have the innate ability to will yourself to perform better, why don’t you do that all the time? The counter point is that clutch shouldn’t really be defined by performing better in tight spots, it should be about not getting worse when the pressure is on. A clutch hitter is someone who maintains the normal faculties in high leverage spots rather than getting worse. This is still an empirical matter, but it’s a more logically consistent one. Seeing a guy hit a home run in the 9th inning makes him appear clutch, but if he struck out four times before that, why didn’t he use his clutch ability to perform better earlier? 4. Pressure Affects Everyone Finally, the idea of clutch is founded on the idea that only the batter is experiencing the big moment. Certainly, no one believes that, but the implication is important. If a clutch hitter faces a clutch pitcher, does that wash out? Same with two non-clutch players. Will an average hitter look clutch if he’s facing a pitcher who can’t throw a good pitch with the crowd roaring? A high leverage plate appearance, if we could define it, is still very dependent on the quality of the pitches being thrown. Even a terrible clutch hitter should be able to crush a batting practice fastball, but a superb clutch hitter still might not touch an Aroldis Chapman fastball. You can control for the quality of the opponent, but not the opponent’s own “clutch” ability. It’s too endogenous to disentangle. ***** The goal here was not to convince you that clutch hitting is a myth, but rather to offer you some resources to explore the issue on your own and some pressing questions to challenge your beliefs. Clutch hitting could exist, but you have to think very carefully about you actually mean when you say a player is clutch and how you have decided that he is. The data generally suggests the role of clutch hitting is limited if it exists at all, but there are more ways to explore the question that could lead us to different conclusions.