Hitting the Brakes on Clutch

As of Sunday, the playoffs have officially begun. The cool-as-a-cucumber consistency required for the regular season (and Round One) now gives way to the hot and heavy mettle of the postseason. Gaudy statlines on a midwinter’s eve in Cleveland aren’t good enough. What matters is playing well when the game - and the season (...and even the franchise) - are on the line. Every playoff minute is important to a degree, but their importance skyrockets in the waning minutes of close games: in other words, clutch time.

It is thus unfortunate that Khris “Khash Money” Middleton, he who taketh the clutch shots, will not play against the Celtics. In his absence, the allegedly unclutch Giannis, the oddly clutch Jrue, and whatever clutchness that role players can summon up from their clutches will likely decide the series.

At least, if clutchness exists.

I’m taking the opportunity of our first real playoffs series, and the notable omission of our most clutch player, to unfold some dirty laundry: I don’t think that players can actually be clutch. I’ll spend the first few paragraphs outlining my definition of clutchness with the help of my good friend Norm, the normal distribution. I’ll then put on my grad student hat and conduct a wafer thin review of the literature on clutchness to see if my skepticism bears out.

When people say clutch, they are referring to a wide variety of things. (as is an evergreen problem of humanity). I’ll try to sketch out a friendly yet clear definition here.

To me, clutchness involves a player playing consistently better than their average in the most important moments of games. There are three main parts to this. I’ll start with the last: the most important moments of games. I acknowledge that people use the word clutch to talk about events that are longer than moments (Game 6 of the NBA Finals, for instance). But I think the majority of our usage of the word stems from more specific moments, typically near the ends of games where teams are within a score or two of each other. The exact contours of “important moments” are subjective, but I’ll defer to the NBA’s definition as moments in the last five minutes of games where teams are within five points.

The other parts are more interesting. Let’s start with a player playing better than their average. This is where Norm comes in. Say hello!

https://www.etsy.com/listing/48582479/standard-normal-distribution-plushie

For some quick background, a lot of things in life take the Norm’s shape. Take height: some people are really tall (Norm’s right-hand side), some people are really short (Norm’s left-hand side), but most people are average height (Norm’s smiling center). Further, the dotted lines are one standard deviation apart from each other. The proportion of folks plus and minus one standard deviation from the mean - in other words, Norm’s smiling center plus his rosy cheeks - is 68%. That leaves 16% remaining on each wing - file that number away for later. Due to these and other desirable properties, statistical models often assume that variables are distributed like Norm.

We can also consider basketball players as falling on a normal distribution. There are some players who are very good, there are some players who are very bad, but most players are average. (Certain basketball statistics are less like this, but we’ll follow the statisticians and stick with Norm.)

Crucially, we can create two normal distributions: one for clutch minutes and one for non-clutch minutes. Another nifty property of Norm’s is that subtracting one normal distribution from the other yields another normal distribution. The result is that there is a normal distribution for the extent to which players outperform their baseline in clutch moments.

Based on my definition, players must play better than average to be considered clutch. Because playing only a teensy bit better than one’s average doesn’t seem worthy of the clutch moniker, I’ll somewhat arbitrarily define the cutoff as the plus one standard deviation discussed above: to the right of Norm’s cheek.

The last part of the definition is the word “consistently.” As reviewed previously in this venue, luck is fickle. Particularly due to the small sample of clutch minutes to serve as a comparison, a player could easily be considered clutch one season but not the next. If clutchness is indeed entirely luck, we would expect each player to be clutch during roughly 16% of the seasons that they play (that number from above: how many players fall above the plus one standard deviation threshold due to chance). Thus, I would like to reserve the title of clutch for those above the threshold significantly more than 16% of the time.

Before diving into the literature on clutchness, I want to quickly discuss two implications of my definition of clutchness, both of which reflect umbrages that I hold dearly. The first is that clutchness is not simply the absence of choking. (For the record, I’m not coming down on whether or not choking exists in this piece.) If a player plays just as well during key moments than they would otherwise, I would argue that they are neither clutch nor chokers. Although I acknowledge that it does take some of the proverbial ice in the veins to maintain equilibrium during these moments, I maintain that clutch players elevate their game when the game is on the line.

The second implication, related to that last point, is that the best players are not necessarily the most clutch players. When we break down clutchness using the pair of normal distributions above, this point becomes clear. It is hard for star players to elevate their games from their already elevated states. Indeed, despite frequently being lauded as clutch, Khris is a three-time all-star who is quite good at basketball during non-clutch minutes. Against the Celtics, the Bucks will probably miss his non-clutch minutes more than they miss his clutch minutes.

Now, to the literature! In an effort to keep work at work, this was a very cursory review. As such, I typed “basketball player performance clutch” into Google Scholar, clicked on some articles, followed some citation trails, and ultimately relied on one primary article that had a spanking literature review itself. One concern is that there are a variety of terms that overlap with clutchness, and my quick review certainly did not untangle that mess.

The conclusion of the review of clutchness in Solomonov et al. (2015) is that, if clutchness exists, it has a small effect and only manifests for a few players. Clutch ability is typically defined as simply any improvement above the mean (e.g., Otten, 2009), dispensing with my more stringent threshold of a standard deviation above the mean. It has typically been studied in the context of baseball, where Bill James himself has questioned its existence.

Influential studies on clutch ability include Cramer (1977), who found no year-to-year reliability in players’ performance in clutch situations (akin to my assertion of consistency). Moreover, in line with my assertion that the best players are not necessarily the most clutch, Otten and Barrett (2013) find that some of the least clutch postseason players in baseball happened to be some of the greatest players of all-time. Players like Mickey Mantle set such a high benchmark during the regular season that a few bad playoff series renders them decisively unclutch.

Much to my chagrin, however, there is actually some evidence that the best players are slightly more clutch. The findings of Solomonov et al. (2015) suggest that, although clutch players (as selected by “experts”) do not show a better uptick in shooting percentage during clutch moments than non-clutch players, they apparently exert greater influence in other facets of the game.

In fact, the authors speculate that clutchness is a buzzword for this venue: a social construct! We remember when players hit clutch shots early in their career and then interpret all subsequent clutch shots accordingly; in other words, we treat clutchness as a dispositional rather than situational factor. Clutch players end up being (only somewhat) clutch because they, their teammates, and their coaches prepare them for clutch moments (e.g., by keeping them out of foul trouble, drawing up plays for them, and so on). If anything, Khris’ absence during clutch moments may be more psychologically damaging.

These findings sync with research by Otten. The idea that great players may actually be more clutch might relate to the importance of perceived control, identified by Otten (2009) as a key component of holding up under pressure. Star players presumably experience perceived control to a greater degree. (Side note: my heart goes out to the undergraduate research assistants who ran this study, which involved 243 participants shooting 30 free throws.) Moreover, in their study of hitting versus pitching in baseball, Otten and Barrett (2013) find that hitters are more subject to pressure, which they interpret in light of the relative difficulty of hitting over pitching. We focus on shot-making in our typical view of clutchness, but that is one of the more difficult parts of basketball; indeed, the clutch basketball players in Solomonov et al. (2015) were not able to elevate it.

Taken together, it appears that clutchness - even under a looser definition that does not rely on Norm’s plus one standard deviation - doesn’t appear to hold much weight. For the most part, my pre-existing beliefs about consistency and clutch players not being the best players held up - woot! Yet, there is some evidence that clutchness exists for some of the best players, at least as a psycho-social effect. More research is needed in basketball and across the landscape of clutch-adjacent verbiage, perhaps using the absence of Khris in this series as a natural experiment.

To conclude, I would be remiss to forget a bibliography. Always cite your sources!

Cramer, R. D. (1977). Do clutch hitters exist?. Baseball Research Journal, 6, 74-79.

Otten, M. (2009). Choking vs. clutch performance: A study of sport performance under pressure. Journal of Sport and Exercise Psychology, 31(5), 583-601. https://doi.org/10.1123/jsep.31.5.583

Otten, M. P., & Barrett, M. E. (2013). Pitching and clutch hitting in Major League Baseball: What 109 years of statistics reveal. Psychology of Sport and Exercise, 14(4), 531-537. https://doi.org/10.1016/j.psychsport.2013.03.003

Solomonov, Y., Avugos, S., & Bar-Eli, M. (2015). Do clutch players win the game? Testing the validity of the clutch player’s reputation in basketball. Psychology of Sport and Exercise, 16, 130-138. https://doi.org/10.1016/j.psychsport.2014.10.004