The One Stain on Kris Bryant’s Record
A few days ago, the Cubs rallied to beat the Blue Jays in 10 innings. It was the 10th inning that was the most dramatic, but the Cubs had a chance to finish things off the frame before. In a tie game in the bottom of the ninth, Kris Bryant batted with two on and one out, against Ryan Tepera. It was one of the highest-leverage plate appearances for Bryant on the year, and he quickly found himself behind 0-and-2. A borderline ball call extended the at-bat, but then Tepera threw a pitch outside. The screenshot says everything you need to know about what happened next.
Kris Bryant is one of the best players in baseball, and he might well be the best player in the National League. Of that, there’s no question, and before we get any deeper, I want to try to get one simple point across. It’s probably futile, but, anyway: There’s a difference between saying a player is unclutch, and saying a player has been unclutch. The former would be a hell of a statement. The latter is easy enough to demonstrate with evidence. Clutch performance tends to be volatile; it hasn’t been shown to be a sticky attribute. It is not my belief that Kris Bryant is actually, naturally, unclutch.
But Kris Bryant has been incredibly unclutch. Historically unclutch. It’s the one place where he’s come up short. As much as I love the things he can do, the data he’s assembled is stunning.
Why don’t I believe that Bryant is unclutch? Well, generally, it would be a silly belief. The starting point for any given player in the major leagues ought to be that said player can handle the spotlight, that he can handle intense pressure. Major leaguers have been selected in part for their immunity to stress. You’d need a lot of evidence to be convinced that someone has this particular weakness. It’s otherwise so difficult to imagine. And then, specifically, Bryant was a clutch hitter a few years ago, as a rookie. He made an immediate positive impression.
So I’m not over on that side of the fence. And yet — and yet, there’s the information we do have. As you might know, our leaderboards track a Clutch statistic. It’s founded upon win probability, and it essentially measures how much a player has actually helped, versus how much you would’ve expected that player to help. It’s a measure of timing, and although it isn’t perfect, nothing is. And this year, among hitters, Kris Bryant has baseball’s very lowest Clutch score. That’s half of this.
The other half is that, last year, among hitters, Bryant had one of baseball’s very lowest Clutch scores. That’s two years in a row, in a statistic that tends to wildly bounce. Single-season Clutch doesn’t mean very much because the sample is small. Two-season Clutch doubles the size of the sample, and this scatter plot, with Bryant in yellow, demonstrates just how much he’s stood out from everyone else:
That’s a plot that would have a flat trend line. That’s a plot that’s basically random, as one would expect. We’re all conditioned to be clutch-baseball skeptics, and for good reason. That doesn’t mean that unclutch performances don’t take place. Over the last two seasons combined, 280 players have batted at least 500 times. Bryant ranks third in WPA/LI, and ninth in RE24. He ranks 40th in actual WPA, because he ranks dead last in Clutch.
We have, on our leaderboards, win-expectancy data stretching back to 1974. That’s more than four decades of information, so I set about examining what it has to say. It’s pretty easy to look up the lowest single-season Clutch marks. I decided to work a little harder, calculating the lowest consecutive-season Clutch marks. Bryant has fared poorly these two years in a row, so, given that, where does he rank within a historical context? This’ll speak for itself.
| Player | Year 1 | Year 2 | Y1 Clutch | Y2 Clutch | Total Clutch |
|---|---|---|---|---|---|
| Kris Bryant | 2016 | 2017 | -2.37 | -2.94 | -5.31 |
| Gary Carter | 1978 | 1979 | -0.98 | -3.73 | -4.71 |
| Gary Carter | 1979 | 1980 | -3.73 | -0.89 | -4.62 |
| Sammy Sosa | 2001 | 2002 | -1.69 | -2.88 | -4.57 |
| Josh Reddick | 2011 | 2012 | -0.62 | -3.83 | -4.45 |
| Tony Armas | 1984 | 1985 | -2.37 | -2.08 | -4.45 |
| Damion Easley | 1999 | 2000 | -2.51 | -1.93 | -4.44 |
| Javy Lopez | 2003 | 2004 | -1.58 | -2.77 | -4.35 |
| Jermaine Dye | 2008 | 2009 | -2.63 | -1.70 | -4.33 |
| Ken Griffey Jr. | 1997 | 1998 | -2.07 | -2.25 | -4.32 |
Last place. Or, I guess, first place, if you like to play with your sorting. From a win-expectancy perspective, over the past two seasons, Kris Bryant as a hitter has been something like five wins less valuable than you might’ve expected him to be. He’s still been helpful, because his abilities and skillset are elite, but this might be literally his one single negative. Bryant has delivered, overall, but he hasn’t delivered in the most important spots. Not with regularity. This table here helps to explain the table just above.
| Leverage | 2016 wRC+ | 2017 wRC+ | Combined wRC+ | Percentile Rank |
|---|---|---|---|---|
| Low | 167 | 181 | 173 | 100% |
| Medium | 144 | 111 | 131 | 87% |
| High | 49 | 24 | 37 | 1% |
In the least important situations, Kris Bryant has been baseball’s second-best hitter. In the most important situations, Kris Bryant has been baseball’s second-worst hitter. I set minimums of 100 plate appearances for each split. The only player worse than Bryant in those important spots has been Ender Inciarte, whose two-year high-leverage wRC+ is an unfathomable 10. That’s an awful lot worse than Bryant’s mark, but still, Bryant is the next player up on the list. Every hit and walk matter, no matter the circumstances, but Bryant hasn’t had that even distribution.
In part, this is probably why the NL Central is even remotely competitive. The Cubs still have the lead, and the Cubs are still the overwhelming favorites, but their lead is just 3.5 games over the Brewers, and 4.5 games over the Cardinals. Offensively speaking, the Cubs rank last in the NL in Clutch. The Cardinals are only a little higher, but the Brewers are higher still. The Cubs as a team haven’t delivered so many of those big hits, with Bryant being the single biggest underachiever. With a few more clutch knocks, the Cubs would’ve been able to more effectively pull away. Perhaps they won’t have needed those hits in the end, but timing has allowed that division to stay a legitimate race.
To get back to where I started, Kris Bryant should still be considered one of baseball’s premier superstars. There are few things he can’t do, and even this one significant issue is almost certainly just a function of noise. When Bryant’s career is up, these seasons will probably look like random blips, signifying nothing but their own joint existence. But one cannot deny they’ve existed. It is, one could safely conclude, the damnedest thing.
Jeff made Lookout Landing a thing, but he does not still write there about the Mariners. He does write here, sometimes about the Mariners, but usually not.


Oh boy these comments are about to become a hot take factory.
Ha. Here’s one:
So I guess playing the Reds, no matter the situation, is considered unclutch. Because that’s where most of his stats come from. Without the Reds last year he doesn’t even sniff MVP.
Perhaps beating the Reds to a pulp should be considered clutch.
Finally someone wrote about this
I think most Cubs fans have been aware of this issue for a while now.
“Issue”
-5.31 in two years is more negative clutch than every CHC player in their entire careers except for … Sammy Sosa, who’s at -14.09 lifetime.
Sammy earned negative clutch in every single season that he was a Cub. I mean, there’s worse company Kris could be in.
As everyone doubtlessly remembers, when this article was written last season Bryant immediately hit two ultra-clutch homers and the “one problem with his MVP case” managed to resolve itself.
Seems like Bryant’s clutch score for all of 2016 was extremely poor, and a case could be made in favor of the considerably-less-unclutch Corey Seager having deserved the MVP.
“This is a thing that might be a thing… or not.”
– Every Jeff Sullivan Article
Which is how knowledge grows from nothing into something.
As a scientist, this is 100% accurate.
I’d like someone to to more analysis on batted balls for Bryant. What I’m wondering is if he is getting unlucky in these high leverage situations? A quick glance at 2017 I see his high leverage BABIP is .156, yet his hard hit % is 28%. Seems to me that 2017 has been mostly unlucky. 2016 looks different at a glance, with a 36.7% strikeout rate.
A couple months back we had an article on how Pujols has been clutch, and oh boy did that regress to the mean, which is an apt demonstration on how prone this stat is to changing.
Also, worth remembering that the guy did the needful in about as clutch-ey a situation as possible, fielding the final grounder in the first world series victory for the Cubs in a century. The guy was grinning the whole play…that response makes me skeptical that he’s some sort of nervous mess in clutch situations.
I’ve read the glossary entries time and time again yet am still unable to easily differentiate “clutch” from “win probability” from “leverage index”.
My probably simplistic understanding:
WPA – the contribution a player has made to the probability the team will win the game
LI – the importance of an event in a game as measured by the potential swing in win probabilities triggered by different outcomes for that event
clutch – a measure of the WPA contributed in high LI situations. In other terms, the extent to which the player contributes positively (WPA) in situations that have a high impact (LI) on the result of the game
It’s a little more complex than that. Clutch is a comparison of WPA to WPA/LI. WPA is the some total of win probability chances of all player events where as WPA/LI first “normalizes” each event be dividing the win probability change by the LI before summing the totals.
So for example: If Kris hit’s a HR in a situation with a leverage of 0.5 for let’s say a 10% increase in win probability and then strikes outs later when the LI is 1.5 for say a 6% decrease in win probability he ends up with a WPA of 0.04 for the game (.1 – .06), but a WPA/LI of 0.16 ( .1/.5 – .06/1.5). We take the difference of the two to determine Kris’ clutch score which in this example is -.12.
One last note for completeness, I purposely chose an example where the average of the LI events was 1. Over the course of a season it’ll always be pretty close to 1, but even so for some players this won’t be the case. For this reason the actual equation for Clutch is WPA/pLI – WPI/LI where pLI is the average LI for all events the player had.
I hope this helps.
What is a clutch situation? Runners in scoring position? While most hitters would be better w risp it’s not a universal truth.
They actually have a formula for it
http://www.fangraphs.com/library/misc/clutch/
My understanding (correct me if I’m wrong) is that the “clutch” statistics mostly measure how you do in “clutch” situations relative to your overall performance. So overall performance is the baseline…which means that guys like Alcides Escobar and Albert Pujols (who are objectively terrible hitters) wind up having great clutch scores and Kris Bryant and Aaron Judge (who are objectively awesome hitters) look worse.
I am just always skeptical of a score where lousy hitters so frequently wind up at the top and excellent hitters so frequently wind up at the bottom.
What you say about Clutch is true, it would be just about impossible for someone like Alcides Escobar to be last in the league in Clutch, simply because he’s such a bad hitter to begin with.
To wit, top/bottom 5 clutch performers last 3 years:
2017
Top: Pujols, Mookie, Mauer, Melky, Arenando
Bottom: Bryant, Judge, Jose R., Inciarte, Peralta
2016
A. Garcia, Galvis (?!), Leonys, Beltre, Alonso
Nelson, Saunders, JBJ, Bryant, Dozier
2015
Hosmer, Moreland, Xander, Prado, Bryant
Semien, Harper, JD Martinez, Machado, Belt
Which Peralta this year?
It seems like maybe when these guys are up in a clutch situation, they are almost certain to face a reliever who’s been brought in specifically to get them out.
I think this is quite likely–a selection artifact of sorts.
I think its still a fluke. Jeff briefly touched on this but he actually had one of the highest WPA in baseball in 2015. In the 2016 playoffs, he had a game tying HR in game 3 of the NLDS, had key HRs in games 5 and 6 of the World Series (albeit in low leverage situations, but in an elimination game in the WS arent they all high leverage situations to some degree?), had 4 hits in game 6 of the WS and overall had a .923 OPS in the 2016 playoffs.
I know I’m cherry picking specific moments where he did come through, but it seems clear that he doesn’t have some sort of mental issue where he can’t perform under pressure.
Are we just ignoring that Bryant’s 2015 Clutch score was 5th highest in baseball?
I think the dramatic shift from 5th highest –>4th lowest–>Lowest is pretty damn interesting, too.
The article explicitly acknowledged Bryant’s excellent 2015 clutch performance.
Where?
“And then, specifically, Bryant was a clutch hitter a few years ago, as a rookie.”
Ahh, thanks. I read thinking “2015”, not “rookie”. I still think the dramatic shift from top 5 to bottom 5 is a bigger story.
Some pretty good names in there
I decided to look at Ken Griffey Jr. two cited years
1997: .304/.382/.646, led the league in R, HR, RBI, SP, and TB. 9.1bWAR and MVP
1998: .284/.385/.611, lead in HR, 6.6 bWAR and 4th in MVP voting
This validates what my gut has told me (despite wild knee jerk condemnations by Twitter every time the issue is raised), but I wonder why. Who would think LaStella has a .450+ BA with bases loaded too. I’m wondering if there is a difference in focus or pitch expectation for clutch players. Kris is SO methodical (according to his father), I wonder if he doesn’t adjust his approach in higher leverage positions. Curioser and Curioser..
As I understand it, Clutch is not generally predictive. It doesn’t correlate well from year to year for most players, and most perform in the clutch much as they do in low leverage situations…over time.
There are outliers, but absent quantifiable explanation it’s hard to say if it’s a skill, or just a pattern that sometimes comes from randomness
I asked about this on an earlier post, and was told that being “clutch” is not a repeatable skill in the sense that you suddenly become a better hitter with runners in scoring position. That was what I was thinking too.
However, it is “repeatable” in the sense that some guys are vulnerable to a shift, and this might change with runners in scoring position. So the “repeatable” part of “clutch” is actually a story about the fielders.
(When told this, my brain exploded)
If true, to the extent that clutch is repeatable it is because you have a serious weakness that becomes less of a weakness with runners in scoring position. And also a huge dose of random error, plus another substantial amount of correlated error with your baseline performance. So being “clutch” doesn’t seem like it’s related to being “clutch” in the colloquial sense of the word.
Great player. Hope he’s a Cub for life. But this year when he’s up in a big spot it’s easy money to bet on a pop up or a K. The quality of his ABs in these spots are generally horrific IMO.
It’s worth noting that Bryant’s career BABIPs in L/M/H-leverage situations are .347/.348./.304. That gives some evidence that this is basically a matter of bad luck.
It seems at least possible, though, given how wild the change was between 2015 as opposed to 2016-7, that Bryant changed something in his hitting tactics between 2015 and 2016. Maybe he started feeling like he had to produce a hit in clutch situations and started swinging at bad pitches? To investigate this it would be necessary to check differences in plate discipline in high-leverage situations in different years. Unfortunately, the tools which Fangraphs at the present time don’t allow this (do they?).
In high leverage situations, Bryant’s LD% is lower, his FB% is higher, his IFFB% is higher, his Soft% is higher, and his Hard% is lower. Hitters have considerable influence over their batting average on balls in play, and all of those factors help explain why Bryant’s is so much worse in “clutch” moments.
Also, 44 points of BABIP is still an appreciable drop. It’s not huge, true, but it’s still present.
Yes but the BABIP drop is explained by higher FB% and IFFB%, both of which correlate with lower BABIP. So it’s not just luck, he’s actually performing differently.
Of course, I assume he’s hitting worse because often good relievers are brought in to face him in high leverage situations. I wish clutch could somehow be weighted for opposing pitcher quality, seems unfair to say a player isn’t clutch just because his competition gets better during the time period Clutch (the stat) is measuring.
This year the average % difference in hard-hit balls for all batters between low-lev and high-lev situations is 3% (lower in hl situations, presumably for the reasons you mentioned) which is exactly Bryant’s career average, so we can’t attribute anything to bad clutch hitting here. For line-drive % it doesn’t change league-wide for leverage, so Bryant’s a little worse here, but it’s just 3%, which is probably not statistically significant. Still looks like just bad luck to me.
I wish I could find more leverage splits, but the information I have been able to dig up makes this look like something real. While it’s not a perfect approximation for leverage, BrooksBaseball does show a sharp increase in Bryant’s whiffs versus fastballs as the game progresses, becoming particularly alarming in the 8th and 9th innings. I would be interesting to know if he was facing much higher fastball velocities in those innings (I assume so) and how fastball location might be changing depending upon the situation.
FanGraphs’ leverage splits do show that Bryant’s HR/FB% in high leverage situations is about a third of its normal rate. His IFFB% is also significantly elevated while his Pull% is drastically reduced and his Soft% is up. There’s too much here to believe that he’s just enduring bad luck. Either Bryant is adjusting his swing/approach in high leverage situations or pitchers are exploiting him by attacking him differently.
This was informative. Does Bryant have unusually poor splits against high-velocity fastballs? If so, that could explain a great deal, given that such pitches tend to be encountered later in games…
I remember after the Dodgers swept the Cubs back in June, both Maddon and Bryant were talking about the problems the Cubs were having with the Dodgers high-fastball strategy.
Its not uncommon for a hitter with a lofted swing like Bryant has to struggle more against pitches up in the zone, and while the Dodgers are a more extreme example, I bet many teams have bullpens that not only throw harder, but throw up in the zone more (since harder throwers often throw more 4 seamers), which is something it wouldn’t be surprising for him to struggle with given his swing mechanics.
If the data are just for innings, that isn’t a good substitute for leverage because the 8th and 9th innings also have the LOWEST leverage situations. HR/FB% can be largely a matter of luck. IFFB% and Soft% wouldn’t be but has to be compared against league averages because of the possible effect of facing better pitchers in high-leverage situations. It would not surprise me if one of the possibilities you mention in the last sentence is the case, but it would have to be demonstrated by statistics which we don’t have. I have to suggested that Fangraphs devotes some energy to giving more leverage splits so that we can check things like this, but this hasn’t happened. I would assume that this is related to the ideological resistance at Fangraphs to any belief in clutch hitting as a real thing, so people there would have no interest in publishing statistics which could be used to demonstrate that clutch hitting exists for anyone.
I think I might have an actual reason: Mr. Bryant has faced extremely weak pitching. In 2016, he faced opponents 11% worse(baseball prospectus) than league average. Most of this would naturally occur vs starting pitching, because logic 🙂 It is possible that when he faced relievers that were much better than these starters, he was blown away. Relievers normally have higher leverage spots, so he would have faced these stingier pitchers in said higher leverage spots, which makes him seem incredibly unclutch. This year he is just as unclutch, and has faced pitchers 7% worse. Summary: He destroys awful SPs and “withers” versus RPs that seem a lot better to him than a player that faces average SP.
Hopefully you could understand that ramble 🙂
ALSO, he has fairly large platoon splits, so the relievers he faced would be even more difficult to bat against.
Oh yeah, the old “he can only hit bad pitching argument”. Been a while since I’ve seen this sizzling take.
On the pitching side:
2017 Top 15:
1) Jordan Zimmerman (R)
2) Kevin Gausman (R)
3) Jason Vargas (L)
4) Drew Pomeranz (L)
5) Gio Gonzalez (L)
6) Clayton Kershaw (L)
7) Alex Cobb (R)
8) Clayton Richard (L)
9) Patrick Corbin (L)
10) Wade Miley (L)
11) Ervin Santana (R)
12) Mike Fiers (R)
13) Ricky Nolasco (R)
14) Mike Foltynewicz (R)
15) Justin Verlander (R)
2017 Bottom 15:
1) Jose Quintana (L)
2) Michael Fulmer (R)
3) Jason Hammel (R)
4) Jeff Samardzijia (R)
5) Mike Leake (R)
6) Jordan Montgomery (L)
7) Tanner Roark (R)
8) Corey Kluber (R)
9) Ian Kennedy (R)
10) Chris Sale (L)
11) Chris Archer (R)
12) Zack Greinke (R)
13) Zach Davies (R)
14) Ariel Miranda (L)
15) Jon Lester (L)
I’m not sure how much stock – if any – to buy into this. You could easily make the argument that the bottom 15 has more collective talent.
(I choose 15 and 15 because the league is comprised of 30 teams)
When the “least clutch” pitchers happen to include basically 5 of the Top 10 pitchers in the league this year (Sale, Kluber, Fulmer, Archer, Greinke) and basically everyone else is above average except Miranda…that says to me something is weird with this statistic. Perhaps this is largely a selection effect, with better starters getting longer leashes to get out of jams or something.
Interestingly, there are a number of really good-[erforming pitchers on the most clutch list (Kershaw, Pomeranz, Vargas) and sprinkled in among a lot of guys who have struggled.
I think for pitchers the data is a lot noisier, especially for starters.
Starters have a really high proportion of low-leverage PAs and need certain circumstances to align to get high-leverage PAs, namely, they have to pitch late into the game, and the score has to be close. If either of those things stop being true before the later innings, a starter won’t get any high leverage PAs in a start. A starter who gives up 2 runs over 7 innings, and gives up the two by giving up a double and a home run with less than 2 out may simply never face a single PA with a LI above 1.5 or so.
This means that the clutch or uncluthchness of starters is probably determined by something less than 50 PAs, even for guys like Kershaw. Kershaw is probably on the clutch list almost exclusively because of one game he pitched, a 9 inning game where he allowed only 1 run in the 9th inning to score on a wild pitch from second base, but where Lance Lynn had also allowed 1 run and the game eventually went 12 innings.
Since he pitched so many PAs late in a close game, he probably got like 1/2 of his ‘high leverage’ PAs this season in that one outing, and hes pitched other complete games and into the 8th inning a few other times.
There’s just not that many opportunities for a starting pitcher to accumulate high leverage ABs, especially in a world with more home runs. Home runs can create 2 or 3 runs in situations that never qualified as high leverage, and then often no other PA in the rest of the game will be high leverage for that pitcher (if you’re trailing by 2+ runs, even RISP situations against you tend to be low leverage because you’re already losing).
I watch the Fangraphs WP charts for games I watch very frequently and I’ve gotten pretty used to what it takes to see a leverage index go above 2 or 2.5. Its not that common. If you look at the Fangraphs Wp graphs, one of the stats it shows is the overall game leverage index for each pitcher. Starting pitchers almost ALWAYS have gmLI under 1.00.
Clutch for starting pitchers is basically meaningless data. I could see it being meaningful for relievers since a much higher percentage of their PAs (and probably a larger actual amount of PAs) are high leverage due to how good relievers are used.
Starting pitchers typically have their highest-leverage situations relatively late in close games and so they do in such situations is affected strongly by whether their managers pull them faster or leave them in longer. Pitchers who get left in longer pitch tired more in high-leverage situations and so naturally have worse Clutch scores.
I don’t now about the others, but Samardzija has serious problems pitching from the stretch so he’s much worse with runners on base which would naturally be in generally higher-leverage situations.
Certain types of hitters will hurt worse w runners in scoring position. It’s rare but it’s due to possibly getting pitched around affecting their approach
This is my guess for the best explanation for bad clutch hitting when it actually exists. Good hitters feel a lot of pressure to produce hits in high-leverage situations as opposed to walking which often means passing the responsibility for getting the big hit on to the next hitter. Jeremy Affeldt made a career out of getting batters to swing at bad pitches in high-leverage situations. The problem is that Fangraphs does not give the statistical tools to see if this is actually happening (hard to imagine this is an accident, it would be like the Catholic Church teaching astronomy in the Middle Ages).
Actually, astronomy was a required core subject for virtually all medieval Catholic schools.
Learn your history, kids, or you’re going to look as stupid as johnforthegiants.
The fact that the bottom of the clutch leaderboards have good players doesn’t mean something “fishy” is going on with the statistic and you can’t trust it. It’s an expected side effect of a statistic that isn’t predictive and fluctuates year to year.
Clutch measures a player against himself. League leaders should be equal parts “good” players and “bad” players, by design. It’s not a bug in the statistic.
Actually, that is not true for a couple of reasons.
First, there is a quite a bit of survivor bias.
If a ‘bad’ player is unclutch, it is more likely he will see his opportunities erode and won’t have enough at bats to qualify.
Second, there tends to be limits as to how bad a MLB player (good enough to stock up enough at bats to qualify) can be.
Bryant is batting .128/.308/.128 for wrc+ 24 in high leverage situations this season.
It really can’t get much worse than that.
However, only a handful of players can hit wrc+ of 180 in low leverage situations like Bryant making his clutch score much worse than
somebody, say who hits for wrc+ of 10 in high leverage but 90 in low leverage.
Therefore, the abundance of ‘good’ batters at the bottom of clutch table is not just a random coincidence.
Just for the record,
in 2015 postsesason, Bryant’s total WPA was -0.156.
(His 2015 regular season Clutch score was 1.85, the fifth highest)
In 2016 postseason, Bryant’s total WPA was 0.432.
(His 2016 regular season Clutch score was -2.37, the fourth lowest)
Also of note, in 2013, Trout had clutch of -2.62 (second lowest).
The next year, Trout had clutch of 1.70 (fifth highest).
“Clutch performance tends to be volatile; it hasn’t been shown to be a sticky attribute. It is not my belief that Kris Bryant is actually, naturally, unclutch.”
And yet here you are with another article about it.
So, what I really want to know more about now is that three year stretch for Gary Carter. -0.98, -3.73, -0.89.