Under Pressure: The Psychology of Clutch

By FanGraphs’ metric, Bryce Harper has been one of the most clutch players in baseball this season. (via Ian D’Andrea)

Everyone will encounter high-pressure situations in their lives, and everyone will find themselves reacting differently. Some of us thrive in such high-stakes situations while others falter. As a whole, major league players are much, much better than the average person at pulling off stunning feats under pressure. But even at this elite level, the variation in outcomes is pretty significant.

Since baseball has a measurement for virtually everything imaginable, there’s a statistic based on WPA and pLI that measures clutch. For reference, here are the most clutch position players with at least 200 plate appearances this season:

2019 Clutch Players

And the position players with the lowest clutch statistics:

2019 Least Clutch Players
Player Clutch
J.D. Martinez -2.48
Jorge Polanco -2.10
Justin Turner -1.94
Mark Canha -1.66
Jorge Soler -1.56
Tommy Pham -1.52
Willson Contreras -1.51
Wil Myers -1.50
Yasiel Puig -1.48
Willy Adames -1.43

It’s worth saying that being more clutch doesn’t necessarily mean being a better player – for example, the 2019 fWARs of the fifth-most-clutch and fifth-least-clutch players are 0.6 and 2.8, respectively. We intuitively understand this idea, but as I was looking at the clutch statistic, a couple of questions stayed at the forefront of my thoughts: Under what situations do some players thrive and others struggle? And, more importantly, why?

I wanted to spend some time in the weeds, so I exported some data from FanGraphs into R. Because R is incredible, it’s fairly straightforward to merge data frames and create new statistics at will. In my enthusiasm to see clutch hitting in action (well, in a spreadsheet), I dug through the differentials of statistics like walk percentage, strikeout percentage, wOBA, and ISO, which I feel provide some indication of player performance in situations of varying pressure levels.

Some players thrive under specific stressful situations, while others do the opposite. For the sake of variation, I’ve included players who don’t crack the top 10 list above; these players’ statistics are, to me, incredibly interesting. But, as per usual, my attention was dragged off to the psychological side of baseball, so we’ll also look at the underlying mechanisms behind choking.

Five Players Who Thrive Under Pressure

Christin Stewart (Detroit Tigers, outfield)

With an fWAR of -0.3 this season, Christin Stewart isn’t in the running for any batting titles. And yet, he’s got a clutch of 0.95, landing in the top 15. These stats seemed to contradict each other — until I reminded myself that clutch refers to improved performance in the face of high-leverage situations and not top-notch playing as a whole. Therefore, the “clutchness” of a player can be assessed from the differentials of his stats; a bases-loaded .300 player is more clutch than his .400 counterpart if the former hits .250 on average and the latter maintains a batting average of .380.

I considered Stewart’s hitting this season in that context. In his 327 PA, he strikes out 24% of the time. But in high leverage situations, his K% is only 6% – that’s an 18% difference, which leads the majors. When comparing low-leverage to high-leverage situations, this discrepancy grows to 21%. I’d say that counts as being pretty clutch, considering that not striking out is one of the best player outcomes – especially when the stakes are high. Stewart’s large K% differential is also reflected in his wOBA, which is .299 as a whole and .393 in high leverage situations. Not only is he striking out much less often, but he’s also getting on base more.

Christian Yelich (Milwaukee Brewers, outfield)

Yelich has hit 39 home runs this season, the same as his 2018 total. Compared to last season, he strikes out less often, walks more, and hits with more power. But even though he’s already a baseball giant in the midst of a phenomenal season, he gets even better under pressure.

Stewart leads major league baseball in K% differential between average and high-leverage situations, but Yelich comes in second with 13%. His strikeout percentage of 19%, down from last year’s 21%, drops even further in high-leverage situations to just 6%. In combination with this drop in strikeout rate, Yelich’s wOBA shoots up from .455 to .670. And he has the second largest ISO differential (0.2) in the majors. His slash stats also increase across the board, from .338/.436/.709 to .536/.606/1.107 – hitting above .500 in high leverage situations feels unsustainable, but if there’s any player who continually defies my expectations, it’s Yelich.

Mookie Betts (Boston Red Sox, outfield)

Betts has a clutch of 1.06, so he’s not far from the top 10 in the table above; I don’t think that statistic is particularly surprising to anyone who’s even somewhat invested in the Red Sox. But, when there are men in scoring position, his batting average drops by .015 and his slugging decreases by .131. These aren’t exactly small numbers. What gives?

You might’ve guessed: he gets walked more. A lot more. In his 104 plate appearances that have had men in scoring position – 21% of all his plate appearances this season – he’s been walked 28% of the time. This is the highest in the majors. But – you might say – considering that his average BB% is already in the top 10, maybe that’s not particularly interesting. I’d beg to differ: when there are men in scoring position, his BB% differential is 13%, the second-largest in the majors behind Maikel Franco. In turn, he has one of the worst ISO differentials, from a somewhat above average 0.216 to just 0.100, the 17th lowest out of 158 players. In average vs. medium leverage situations, his BB% differential stays in the top 10, and his wOBA increases from 0.375 to 0.418. Betts gets on base more when it matters most. Talk about plate discipline in action.

Charlie Blackmon (Colorado Rockies, outfield)

You know Charlie Blackmon can hit – after all, you don’t get a .396 wOBA by striking out. (Fun fact: his K% of 18% equals that of Mike Trout and Vladimir Guerrero Jr.) But, when one of his teammates stands on second or third base, Blackmon’s performance improves. His already impressive .396 shoots up to .509, which leads the majors by a margin of .024.

How does Blackmon get on base? He does get walked ever so slightly more, but I think it’s mostly due to making contact. When there are men in scoring position, his strikeout rate drops from 18% to 10%. This K% differential is the ninth highest in the majors; as a result of this drop, his strikeout rate with men in scoring is in the lowest decile. Like Mookie Betts, Blackmon has a knack for not making outs when there are runs to be scored, but unlike Betts, he does so by slapping out hits, not by drawing walks.

Jose Iglesias (Cincinnati Reds, shortstop)

Unlike the other four, he’s not an outfielder! But, along with the support he provides as a shortstop, Jose Iglesias also hits particularly well in high leverage situations – as in, his batting average increases from 0.280 to 0.429, the 4th-highest overall.

What I thought was interesting, though, is how consistently Iglesias’ stats improve in high leverage situations. His strikeout rate drops by 7%, his walk rate increases by 7%, his ISO increases by .175, and his wOBA increases by .208. His slash stats also look like they belong to different players: his OPS in low leverage situations (.657) is lower than his high leverage SLG (.686). It seems like thus far this season, Iglesias has performed significantly better across the board when the pressure is cranked up — an anomaly in the major leagues. Going forward, I’m curious as to how he’ll do in August; as part of the NL Central, it seems likely that Iglesias will face multiple high leverage situations in the near future. And as the stakes continue to rise, will he perform even better?

Five Players Who Struggle Under Pressure

J.D. Martinez (Boston Red Sox, outfield)

I think Martinez exemplifies the difference between a good player and a clutch player – his slash stats are well above average, and his fWAR is 2.2. These are definitely not embarrassing numbers, and yet, he’s got the lowest clutch in the majors — and not by just a tiny bit; his clutch of -2.48 is .38 lower than second-place Jorge Polanco’s.

Martinez’s most significant discrepancy is his K% differential in high leverage situations vs. average. As a whole, Martinez’s K% is 20%, which is fairly average. However, he strikes out in 34% of his high-leverage plate appearances, which is in the bottom 15. Likely because he strikes out significantly more often, Martinez also has one of the largest negative wOBA differentials (-.143). And if we compare low leverage and high leverage situations, these differentials become even more drastic relative to other players: Martinez leads the majors in the worst differentials for K% (.16), ISO (-.268), and wOBA (-.183).

Cody Bellinger (Los Angeles Dodgers, outfield)

Like Martinez, Bellinger is a great player. Actually, great might be a severe understatement. But, though Bellinger can definitely swing a bat (and catch a ball), he’s nearly in the bottom 20 for clutch.

Most Dodgers fans rejoice when Bellinger comes up to the plate in stressful situations, especially when there are men in scoring, but perhaps we should adjust this mindset. Bellinger has the third-highest wOBA in the league (.440), right after Yelich and Trout. However, his wOBA when there are men in scoring is .326, which is somewhat below average. His teammates Justin Turner (.334), Alex Verdugo (.338), Corey Seager (.366), and Joc Pederson (.405) all have higher wOBAs in that situation. Also in that same situation, his ISO differential of -.16 is the third-lowest in the league. Put it this way: if the NL MVP heavily weighted clutch, the clutch Yelich would surge ahead.

Willy Adames (Tampa Bay Rays, shortstop/second base)

An interesting piece of useless trivia: Adames seems to have reversed his strengths over the past season. Last year, his offense was 4.6 and his defense was -2.3; now, his offense is -4.3 and his defense is 5.8. Despite this defensive prowess and his 2019 fWAR of 1.5, Adames has a clutch of -1.43. Though that puts him in the table of least clutch players, I wanted to include him here again because of how often his name popped up at the bottom of performance differentials during my quest to “see” clutch.

Adames gets struck out 11% more in high leverage plate appearances, for a 40% strikeout rate – the fifth-lowest overall. And, for what it’s worth, he hasn’t been walked in any of his high leverage plate appearances this season. Striking out more and getting walked less often has resulted in a significantly lower wOBA as well – his wOBA differential of -.159 is the second-lowest. This difference grows to -.182 for low vs. high leverage comparisons (also the second-lowest in the league).

Pete Alonso (New York Mets, first base)

Even though he’s still in his rookie season, Alonso’s shaping up as a key component of the Mets’ lineup. What stood out most to me is that when the stakes rise – average vs. medium and high leverages, as well as when there are men in scoring position – Alonso gets walked much less often. In all of these situations, Alonso’s BB% differential places him among the bottom of the league.

That got me thinking about where his decreased BB% differential would be reflected. We know Alonso can hit a ball (hello, Home Run Derby), so I checked to see whether his wOBA would go up and also whether his K% would decrease. No such luck: in high leverage situations, Alonso’s K% differential is 3%, which is just slightly more than average. However, his wOBA drops from a stellar .394 to a below-average .283. This differential (-.111) is among the top ten largest in the league. And – for a sanity check – Alonso’s slash stats all decrease in high leverage situations, dropping from .260/.363/.601 to .190/.234/.476. As much as I was hoping otherwise, it seems that Alonso’s performance under pressure doesn’t quite live up to his usual stuff.

Jonathan Schoop (Minnesota Twins, second base)

When I was making these lists, I had a tough time deciding who to put in this slot. Several guys were in the running, but in the end, I decided to go with Schoop because he’s the middle ground of much of what I’ve been trying to emphasize.

Schoop is, on average, a solid player. His wRC+ (95) is offset by his skills at second base, coming together for an fWAR of 1.0. His ISO (.202) also hovers around the major league average, but in high leverage situations, that measurement drops sharply to just .036, putting him in the bottom 10. As a consequence of this ISO differential, his wOBA differential (-.133) is sixth from the bottom of the league. Schoop falls off somewhat in his high-leverage plate appearances, but then again, so do some of the best players in the game – even Mike Trout strikes out more often in high leverage PAs.

What Causes Choking?

I want to reiterate that being a clutch player is not synonymous with being a good player. For example, Bellinger’s been absolutely phenomenal this season, but his clutch is still far below average. That raises the question: Why do some players step up in high-pressure situations while others don’t?

As of late, more psychological research has been directed toward examining high-pressure performance such as standardized testing and job interviews. In Sian Beilock’s Choke, the former psychology professor at the University of Chicago and current president of Barnard College discusses causes of the infamous choke, as well as some tactics that can help overcome it.

Beilock’s explanation of why we choke centers around overthinking: by overcrowding your working memory – the part of your memory that holds and processes information for short periods of time – you become hyperaware of all that could go wrong, and in turn, you become acutely conscious of your own motions. In Choke, she gives the classic example of shooting at the end of a close basketball game. Because the player is so attuned to the pressure of his/her team, coaches, and audience, the player tries to monitor every aspect of a highly familiar action – shooting the ball – and misses the shot.

This example embodies her discussion on how overthinking can disrupt the motor cortex (a portion of the cerebral cortex that controls voluntary motion), especially in the execution of intuitive actions. A similar – though perhaps more surprising – finding was discovered during a study that asked people to describe how they walk. When they were asked to dissect walking, an action that most people don’t often think about, the study’s participants walked more slowly. You can imagine how overthinking in baseball – a game that’s determined by milliseconds – can be disastrous to a player.

Overthinking is also referred to as paralysis by analysis. Experts (Beilock’s book focuses on golf, but we can extrapolate to baseball) perform worse when they are asked by experimenters to carefully consider every aspect of their putt. Over the years, many golfers have performed miserably on the last day of a tournament, dashing away their chance to win. The main takeaway I got from Choke is that choking is largely a byproduct of overthinking – somewhat counterintuitively, thinking about your performance (unless you’re a complete beginner) can result in far worse results than if you’d approached the situation with an empty mind.

But when you’re under pressure, it’s incredibly challenging to not think about what’s at stake. Even now, if I ask you to not think about a purple shirt, you’ll probably struggle to do so. Luckily, Beilock also touches upon a variety of strategies that can be implemented to prevent overthinking – and, in turn, prevent choking.

These strategies largely fall into two categories: those that you do as preparation, and those that you do in the moment. For the former, Beilock’s primary recommendation is to practice under pressure. In some cases, like setting a timer for your practice test, simulating pressure isn’t too difficult. In baseball, however, players play nearly every day during the regular season. In lieu of practice, they’re expected to perform night after night. Maybe practicing under pressure could be further emphasized during spring training, such as by starting batters at pitcher-friendly counts like 1-2, or even by reducing strike counts.

Another one of Beilock’s suggestions is to practice mindfulness by means of meditation or writing down your pressures/anxieties. These practices are widespread in the worlds of golf and shooting, but to my knowledge, they haven’t permeated baseball to a comparable extent.

In the second category – strategies for the heat of the moment – Beilock offers a variety of ideas that are meant to distract yourself. These include counting backward by three, singing to yourself, and focusing on a single word instead of an outcome, like “smooth” for a golf swing. By engaging your working memory in something other than the situation at hand, Beilock explains, the motor cortex will have a greater capacity to perform, executing a motion just as it has been successfully practiced thousands of times in situations when it didn’t really count.

As the science of player development continues to advance, it’s worth wondering if the differences between the most and least clutch players can be bridged. Can players work on their pressure management just as they might work on their swing? How can organizations further help players, especially in the postseason? And, if greater focus were placed on preventing choking, how would that affect individual and team performance?


Miriam Zuo is a die-hard Astros fan who also likes learning about social sciences through books and podcasts. She tweets about baseball @miriamzuo.

15
Leave a Reply

Please Login to comment
11 Comment threads
4 Thread replies
0 Followers
 
Most reacted comment
Hottest comment thread
12 Comment authors
tzRyan BrockJohnstonMatthewsbf21 Recent comment authors
newest oldest most voted
martyvan90
Member
Member
martyvan90

Miriam, thanks for touching the third rail. Most data-driven folks who follow baseball are skeptical of the concept of “clutch”. Traditionally it’s been a bar stool term of fans advocating for their childhood idol who should be in the hall based on their “game against ______ in a pressure situation”. The limits of probability aside, SSS negates the majority of “clutch” narratives.

channelclemente
Member

Yet they have no trouble accepting a correlation in which models explain 30-40% of the variance in a comparison as well as concluding every model is a linear one.

Richie
Member
Richie

Wet blanket here. This could’ve benefitted from first looking at previous research – of which there’s plenty – on whether ‘clutch’ actually exists in baseball, or if it’s instead Mr. Random. Of course, then you might’ve not done the article, as said previous research really, really leans toward ‘it really doesn’t’.

ryanredsox
Member
ryanredsox

“Clutch players” (very likely) don’t exist. There’s been extensive research showing “clutch” is basically random.

Marc Schneider
Member
Marc Schneider

I’m skeptical of her notion that a player who performs better than normal under clutch situations is necessarily a more clutch player than someone who does less well. I just don’t buy her premise that the normal .250 hitter who hits .300 in clutch situations is a more “clutch” player than the .300 hitter that hits .275 in the clutch. It seems to me that this presents issues of sample size (the better hitter is likely to come to bat in more clutch situations) and pitching strategy (the pitcher might be less careful and more aggressive with the lesser hitter).… Read more »

channelclemente
Member

I think the low leverage/high leverage performance ecology is an interesting one to explore.

WARrior
Member
Member
WARrior

It’s pretty well-established that clutch is usually random, I.e., not repeatable. Yelich has a positive clutch rating this year, but it was negative the previous two years, and is negative for his career. Betts is positive this year, but was negative in four of the previous five seasons. Bellinger is negative this year, and was last year, but was positive as a rookie. His wOBA with men on base or RISP was higher than with bases empty. I’ve never understood the concept of clutch. If a player performs better under pressure (which is not only with men on base, but… Read more »

WARrior
Member
Member
WARrior

I’ll add: Eric Hosmer leads all active players with a 5.96 career clutch. That may seem like a lot, but during that period there have been > 100 hitter-seasons with a clutch value of > 1.0. If it were repeatable, one would expect to see much higher career values. Hosmer has the second highest clutch season of all active players, but he doesn’t have another season in the top 30. Only one player makes the top 30 twice, Pujols (1st and 30th), probably because he’s played so long. Trout has the 20th highest season, but another season I think he… Read more »

ice_hawk10
Member
ice_hawk10

“clutch” always makes for an interesting discussion. as human beings we are wired to be really, really good at detecting patterns. unfortunately we often also see them when they aren’t there, and i think that’s where much of the labeling of players as clutch or unclutch comes from. i think pretty much any person can probably think of moments in their lives (professional, personal, athletic) where they have choked, and most likely if you looked at peoples’ brains in that moment i don’t doubt that you would find some funky chemistry going on there not present in normal instances of… Read more »

tz
Member

I wish that the author had used the points you just made as a segue between the statistical and psychological sections of the article. Without that, the ending sounded like a prescription for a non-existent ailment. Do we really think the Dodgers would even consider asking Bellinger to seek special help to keep him from “choking”?

Matthew
Member
Member
Matthew

The fallacy here is that because you are comparing players’ performance in a small subset of PAs (“clutch PAs”) to their overall performances, there will always be a distribution – some players being better, some being the same, and some being worse – in “clutch” situations, simply because the statistics of normal, random variation demands that that will be true. To prove that “clutch” exists, you’d have to prove that there is a “bias” in the overall population of players that is different from simple normal variation. Then you’d have to observe that some specific players are consistently “clutch” or… Read more »

Matthew
Member
Member
Matthew

Put another way, you could substitute another dataset – “PAs on Tuesdays and Thursdays” – for “clutch PAs” and calculate a “TT” statistic using a methodology the same way “clutch” is defined. You’d find that “TT” likely varies across the population of all baseball players just like “clutch,” but we would not conclude that hitting specifically on Tuesdays and Thursdays is a skill, although some players will have hit better on those days than they do overall.

sbf21
Member
Member
sbf21

If a players strikeout rate drops from 24 percent to 6 percent that isn’t a drop of 18 percent. That’s a drop of 75 percent.

Johnston
Member
Johnston

Not this nonsense again.

“Clutch” is not a skill or a talent, it’s just an artifact of randomness. The proof is that it doesn’t correlate from season to season.

“Clutch” is a horrible concept that was disproven years ago.

Ryan Brock
Member
Member

Walk-off grand slam on the day of publishing for the cover player on this article has to feel good.