One of the most heated debates among baseball fans today is over the existence and importance of clutch hitting. I get why it’s a controversial issue. On the one hand, the history of our game is written by baseball heroes; those great players who rose to the occasion when their team’s needed a hit to win or a homerun to stay alive, and delivered in the clutch. Hitting well when it matters most is dramatic, it’s fun, and it sticks in our memories for our whole lives. Really, a lot of how we experience the game comes down to clutch hits.

Further, clutch hits gets a lot of attention because they’re easy for commentators to talk about. It’s simple and reductive: when a player strikes out late in the game with the tying run on third, it’s easy to talk about the game as if it all came down to that one play. In almost every game you can hear a broadcaster talk about a player or team’s ability to hit in key situations, so of course it seems important, especially as compared to something like a player’s line drive rate, which is almost never discussed on air.

But on the other hand, when fans and pros alike have looked into clutch hitting from a more factual and less emotional standpoint, they haven’t found much clutchiness. That is, the great players didn’t seem to be any more great when the game was on the line than other times. Not only that, but players that had a lot of big hits one year, frequently didn’t have as many the next, or maybe ever again. Being good in the clutch doesn’t look so much like a skill, the way making contact, hitting for power, or taking walks does, as like random variation, like a player hitting well on the 14th of every month.

I think it’s a fascinating issue because it raises some good questions. First, what does it mean to be a “clutch hitter?” Second, is it worthwhile to compare a player to themselves rather than other players? Finally, if hitting in the clutch is a skill, how much is it worth compared to other skills? I’m not going to try to answer these questions definitively (maybe they can’t be), but I’m going to address them in an attempt to add some clarity to a debate that’s sure to go on for another decade or ten.

What is Clutch?

The most common way that you’ll hear the Reds broadcasters talk about clutch hitting is in terms of Batting Average with Runners in Scoring position, or BA w/RISP. It’s a shame, because of all the ways you could think about clutch hitting this is one of the worst. Batting average tells you so little, and having guys on second and/or third means very different things in different games. If The Reds are down 4 runs in first inning, there are two outs with a guy on second, and Schumaker lines a single to left field failing to score the runner, was that a clutch hit? No, in fact it’s a nearly meaningless hit all things considered. When you hear people talk about BA w/RISP, your immediate reaction should be skepticism. It is a statistic with so much noise in it as to tell you basically nothing of any value.

Another way to look at clutch is percentage of runners driven in. This is better than just RBI (another measure of clutch), and certainly better than BA w/RISP, but it limits the idea of clutch hitting to driving in runs. If it’s the bottom of the ninth, nobody out, Reds down by one and Votto doubles to lead off the inning, that’s pretty clutch right? A bunt and a sac fly and you’re tied without even needing another clutch hit. Freeing the idea of clutch from simply driving in runs is another good step, and brings in the concept of “leverage.” This is the idea that there are times in a game that matter more than others.

The leverage of a game situation can be measured pretty precisely now, because baseball stat people have measured the probability that a team will win in every possible game situation (you can see the win probability charts in your Titanic Struggle Recaps here each day), and we know that some situations can lead to much bigger changes in win probability. Generally, four things dictate the leverage of a game situations: how late in the game it is, the score, the number of outs in the inning, and the runners on base. The beginning of the game (first inning, no outs, nobody on, tie game) is a lower leverage situation than starting the bottom of the 9th, because the team has fewer outs to go until the end of the game, giving them less chances to score.

Understanding the difference between high leverage and low leverage situations gives a much better measure of clutch hitting. Getting a hit in a high-leverage situation is clutch, and getting an extra-base hit in an extremely high-leverage situation is clutcher. Fangraphs has developed a stat (simply called “clutch”) that tells whether a player performs better in high-leverage situations than they do in neutral ones. In their description of the stat (see the writers note that a clutch score of .5 is above average, and a score of 1 or better is excellent, with that 1 representing 1 game that the player had essentially won for his team simply by performing better than expected in high leverage situations. The opposite is also true, that a player could be unclutch (-.5 being generally less clutch than average).

Is Being Clutch Good?

There is a general assumption in traditional baseball thinking that some hitters are better in clutch situations, and that these hitters should be commended for this. This has always rubbed me the wrong way. When looking at whether hitters are clutch, you’re comparing them to themselves in other situations. If someone does well in high-leverage situations, and we assume that this is a skill, don’t we have to ask, why don’t they chose to do that well in all situations? Are they lazy? If no one’s on base in a blow out, are they giving away at-bats? If so, isn’t that sort of selfish? What exactly are they saving their good hitting for? Do they have a limit of homers they are going to hit on the year and they’re saving them?

As you see, you get down a rabbit hole pretty quickly. Thinking about anyone trying to be clutch on purpose and succeeding doesn’t really say good things about them, because we expect that pro ballplayers should be trying to be their best all the time. The only way that being clutch would be a laudatory thing to be would be if it was unconscious. If the hitter had no choice about it, but somehow, unbeknownst to them, they happened to get more of their hits in big situations, that would probably be good. It’s hard to compliment them on something that they aren’t really trying to do, but all the same, they would probably help your team win a little more.

So I looked at all of the players since 2000 that have had at least 5000 plate appearances to see if there were any that were clutchy. I chose that cutoff because it’s pretty safe to say that these 134 hitters have enough of a track record that their clutchiness should be visible, and that they are pretty good hitters since teams kept running them out there. I identified 10 players that had an average clutch score of less than -.5, and these are really the only players I would feel even reasonably safe labeling as unclutch, despite what many Reds fans and broadcasters have argued. None of them are Reds, and one of them is Alex Rodriguez, and that makes me a little happy.

In contrast, I identified only one player that had an average clutch score higher than .5. To me, this makes him the only player in this millennium that I feel good about calling a clutch hitter. His name? Randy Winn. His yearly clutch score: .65. In second place was Ryan Howard at .47, so there’s a pretty big gap between Winn and his nearest rival for the clutch crown. As noted above, I’m not saying I think Winn tried to be clutch, or that anyone should try to be clutch, but it seems like there’s a reasonable chance that in some subconscious way, Randy Winn really couldn’t help himself and came through in the clutch.

What’s Clutch Worth?

To summarize, getting a clutch hit can make you a hero in the hearts of your team’s fans, but being a clutch hitter a) has very little to do with the stats broadcasters throw around like BA w/ RISP, b) wouldn’t really be a good attribute to have if it was something you were doing on purpose, and c) means that you’re Randy Winn.

But, if one of our young prospects, say Jesse Winker, happened to be the next Randy Winn in terms of his clutch-factor, what would that be worth? Winn enjoyed a 13 year career, and boasted a .284/.343./.416 line in the heart of the Steroid Era. He is 69th out of the 134 player sample in career Wins Above Replacement, a stat that does a good job at looking at overall value, but does not take into account clutchiness. Would you rather have the next Randy Winn than the next Pujols, Mauer, Jeter, Rolen, Ortiz, Cabrera, etc, etc, etc (all players on the list with more career WAR but less clutch)? I personally would not. Now, would you rather have him than the next Maglio Ordonez? That’s where it gets interesting. Ordonez was basically not at all clutch or uncluch, and had a slightly higher career WAR. Perhaps Winn’s clutch-factor gives him that extra umph, and puts him over the top. But really, that’s about all that clutch hitting can do over a career, and that’s for the rare guy who’s getting it done and not even meaning to.

I hope that this helps to clear up the issue of clutch hitting, just a little. I know that the debate will rage on, old-school vs. new and all that, but even the most ardent supporter of the value of good clutch hitting can’t claim that Randy Winn was really that great of a player, and he’s been the best clutch hitter of the millennium by a mile. Rather than focusing so much on what players do in certain specific situations, RISP, RISP with two outs, RISP in the day with one out and men on the corners, etc. it’s clearly better to just get good players and applaud them when they do well. Even though A-Rod has been one of the 10 least clutch hitters in the game, if it were the year 2000 I’d take him over Winn in a heartbeat. Someone should ask Thom Brennaman if he would.

48 Responses

  1. Kurt Frost

    How is this still an argument? I don’t want a guy on my team who only tries with runners on base and mails it in the rest of the time.

    • Gaffer

      I think most people do not get that each AB is different, and how a manager chooses to pitch a player is influenced by the other batters, who is on the bases, and the handed ness match ups (and thousands of other things). It makes sense that Randy Winn (a true journeyman player that mostly batted lead off or also very low in the order) would have been given the opportunity to hit better pitches with runners on, but pitched to more carefully other times. He was never the one player teams “pitched around”.

    • Lord Oracle (@LordOracle22)

      i just think u guys try to take the human element out when talking about things like being clutch or closing. numbers may tell us a lot but they dont tell us everything there is still the human element were not talking about machines here

      • jdx19

        If there is a human element, show me a clutch player in a statistically significant sample size that cannot be explained by regular old chance.

        I don’t think folks ‘remove’ it, we just show that it doesn’t really account for much.

  2. Gaffer

    Do this calculation please for the most “clutch” Red of all time, Tony Perez. Heck, that clutchiness was his main claim to fame and the best argument for HOF induction. Let’s see!

    • Jeff Adams

      Looks as though you and I had the same idea at the same time. Great minds???? Lol

    • Craig Z

      The data on Fangraphs for Win Probability only goes back to 1974. Perez’s high Clutch for a season is 1.47 in 1977 and low is -0.85 in 1980. In 1975 it was -0.83 and in 1976 it was -0.84. The opposite of what you would expect.

      • jdx19

        That makes sense then why I did not get older players in my 1950-2015 search!

    • jdx19

      Top 11-ish (hahah) Reds by FG ‘Clutch’ since 1950, Min 2000 PAs.

      1) Pete Rose (5.60)
      2) Dave Concepcion (4.44)
      3) Dave Parker
      4) Dan Driessen
      5) Ron Oester
      6) Ken Griffey Sr
      7) Eric Davis
      8) Aaron Boone
      9) Shawn Casey
      10) Paul O’Neill (2.21)
      11) Joey Votto (1.99)

      16) Tony Perez (0.33) (Perhaps not as clutch as expected)

      • jdx19

        Based on what Craig Z said above, this may not be the best list, since earlier data from the BRM guys isn’t available (pre-1974).

  3. Jeff Adams

    Would love to know Tony Perez’s “clutch factor”. I recall so many games when Doggie would come up to bat and MB always saying how “clutch” he was.

    • Chuck Schick

      Perez was a great player on a great team whose accomplishments should be celebrated.

      If we consider “clutch” largely about driving in runs, its fair to note that he very likely had far more opportunity to drive in runs than a similar player on a less stellar team…he also had quality hitters batting behind him so he wasn’t pitched around as much as a similar player on a less stellar team. To his credit, he capitalized on the opportunity…..but he had far more opportunity.

      During his peak years, Rose and or Morgan were on base close to 40% of the time. He had players such as Bench, Foster and May batting directly in front or behind him. While I have no way to prove this, its possible that he had one of the highest Runners on Base/PlateAppearance ratios ever.

      • Gaffer

        That was my point as well. But, even in the opposite case being Votto who has had no real protection, and who the Brenamens think is not clutch, he is in the top 10 of recent Reds.

  4. bhrubin1

    So, as someone who is fairly conversant with statistical analysis in a theoretical way, but not at all in terms of the math involved, my question would be, and it’s a real question (I’m not sure of the answer): What is the statistical probability that out of a random sample of 134 players, one of them would have a clutch score of .65 or higher. In other words, I get that it would be unlikely for a specific player to have that high a clutch score for a career simply by chance, but in the context of the whole sample, is this remarkable, or is this a sample size where we would expect to have about n outliers of about x standard deviations from the mean and Winn just happens to be that guy? Would love to get an answer from someone on this site familiar with the math.

    • jdx19

      I’m on it. I’ll have to try and replicate Jeremy’s sample so I can get the correct input parameters.

    • Steve Mancuso

      Yes, great point. You would expect the deviations between how a hitter bats in regular AB and clutch AB to be distributed normally, on a curve. Just like you would expect there to be a normal distribution of how hitter batted on Friday.

      We could find the Randy Winn equivalent for batting on Fridays and call him Mr. Friday. But that would be silly on its face, right? No meaning at all. That player would just be the one (someone has to be the outlier) who happened by chance to hit the best on Friday out of thousands of players on a normal bell-shaped curve.

    • jdx19

      So, I replicated Jeremy’s sample but chose a slightly different way to caluclate average Clutch per season. I used a standard 600 PA ‘season’ as a basis, so I divided each player’s total PA’s in the 2000-2015 sample by 600 and came out with how many ‘600 PA Seasons’ this player had. It didn’t chance the results a ton, but moved Benji and Yadier Molina into 2nd and 3rd behind Winn and dropped Ryan Howard to 5th behind another baseball luminary, Ray Durham.

      So, with Randy Winn’s new number of .6199, I calculated the standard deviation of the sample (.3238) and the mean (-0.09) of the sample and used z-table approximation to deduce that there is approximately a 1.28% chance that a player would have a Clutch score equal to Winn’s in this sample. This, of course, all assumes that the sample is defined correctly by a normal distribution (which is somewhat reasonable).

      So, to tie it all back around, I’d say this quick study helps to show that Clutch isn’t ‘a thing.’ Given a big sample, someone is likely to be ‘clutch.’ And Randy Winn was that guy.

      • jdx19

        No joke, I just used that exercise as a learning experience for my summer intern that sits behind me. He’s not a baseball fan, but he appreciated the problem!

      • Jeremy Conley

        Good to hear. And pretty interesting that by common measures of statistical significance, Randy Winn (the man, the myth, the legend) would be considered solidly clutch. Definitely a good thing to point out to an intern learning about confidence intervals and the like.

    • Jeremy Conley

      Great comment BHRUBIN1, I wondered if someone would ask about this, or provided the info JDX did below, and sure enough Redleg Nation comes through again.

      You’re dead on, and the only reason I didn’t discuss this issue is that, while it was a little tongue in cheek, I wanted to be able to discuss how funny it was that Randy WInn was the only player to cross the (arbitrary) threshold that fangraphs set up.

  5. Yippee

    Barry Larkin was pretty “clutch” as well, at least it seemed that way.

    • jdx19

      Larkin’s career FG “Clutch” is -0.89. So, not clutch (my this measure).

  6. Matt WI

    The World Wide Leader also contributes to this nonsense on the ticker sometimes… you’ll see “Player X struck out with runner on second in 9th” in a close game score sometimes.

    Funny, they never put down “Player Y GIDP in the 3rd inning when the team was down 2-0 and ended up losing 4-3.”

    • Matt WI

      … And while I understand that the earlier in the game, the less leverage there is to the situation (in theory), it isn’t unimportant and sometimes represents the best opportunity a team was going to have. Which then leads to more thoughts about “trying harder” in the 8th inning vs the 3rd. Stupid.

      • Jeremy Conley

        Absolutely, that’s why I discussed the different factors that go into leverage. A plate appearance with the bases loaded is always higher leverage than one with the bases empty, because the potential change in win probability. If you’re down 1 run in the 3rd, bases loaded, 2 outs, and you strike out, your win probability goes down a little, but if you hit a HR, it jumps up a mile. Not as much as it would in the 8th, but a lot. And it could be the highest leverage situation you have for the rest of the game, because you might not have the chance to swing the game’s outcome that much later on.

  7. Steve Mancuso

    Think it’s worth remembering that the Clutch statistic doesn’t mean anything more than batting on Wednesday or Friday. Here’s how FanGraphs describes their stat:

    “Clutch does a good job of describing the past, but it does very little towards predicting the future. Simply because one player was clutch at one point does not mean they will continue to perform well in high-leverage situations (and vice versa). Very few players have the ability to be consistently clutch over the course of their careers, and choking in one season does not beget the same in the future.”

    We shouldn’t obsess (or even care a bit) about differences in Clutch measurement.

    • jdx19

      Agreed. Also, it is important to remember that Clutch is a player being compared to himself. So, saying the Player X with Clutch=0.50 is more clutch than Player Y with Clutch=0.40 is not mathematically accurate, since they have encountered likely very different situations to arrive at their clutch scores.

    • whodeythinkgonnabeatthemredlegs

      Someone started touching on it below, but aren’t the psychological influences that go into an at bat in a high leverage situation vs s Wednesday different? Isn’t that what we are trying to measure?

      • Steve Mancuso

        The data doesn’t show clutch variance any greater than Wednesday variance.

      • Gaffer

        Bet he is on the Reds next year if Walt is still here. Former Cardinal, check. Doesn’t walk, check. Strikes out a lot, check. Meaningless sense of being a “professional hitter”, check.

    • memgrizz

      He was precisely the example I was thinking about when reading the post. Of course, the vast majority of his “clutch” was due to his exposure to The Cardinal Way and BFIB. It’s the only explanation for his precipitous decline once he got to Boston.

  8. streamer88

    Awesome article. As a former pitcher (and hitter, and current golfer, actually) I do strongly believe in clutch performance. I think this is explained by high pressure/nervousness negatively impacting fine motor performance and muscle memory, and it does so at a different amount for each hitter, and each pitcher. Remember, clutch situations are not just high pressure for the hitter, but for the pitcher also. Therefore, if the negative impact is higher for the hitter than the pitcher, then the pitcher will make less mistakes, and the hitter will make more mistakes with his fine motor movements (and decision-making, he has to decide whether to swing as well). Randy Wynn was negatively impacted by high pressure situations (like we all are), but just much less so (on average) relative *to the pitcher he was facing* in that moment, compared to other hitters. This idea is not disproven by those who hit better than their career average in the clutch, as in the clutch they’re facing pitchers who also are attempting to perform under pressure (but doing so much less effectively!).

    This is my position for many fine motor sports clutch arguments (Kobe vs Lebron, golfers in majors, etc). If you’ve competed in these situations, you know that feeling, and you know the battle you fight with it in those moments.

    • Steve Mancuso

      The data doesn’t bear that out, at least for professional baseball players. Players don’t sustain clutch-performance variance over their careers, at least not more than random distribution would produce.

      I’m not disputing your experiences. I got nervous in varying degrees when I played baseball, too. The theory is that pro players are substantially immune to the pressure/nervousness that bother athletes at lower levels. By the time you make it to the major leagues, those ebbs and flows aren’t a factor. Certain players perform well in clutch situations than others – like Kobe, Jordan etc. But those players perform better than other players at all times, not just in the clutch.

    • Matt WI

      I like the reminder that the pitcher could be impacted too, but I don’t think we can safely or correctly infer that whoever “wins” the at-bat is better in the clutch or that the loser was impacted more than the other. Given that even crappy pitchers “win” over 70% of the time (thinking of BA Against), and that I’m sure we’d the larger trends hold in clutch situations.

      Pedro Villarreal had an awesome moment getting out of jam the other night, other times he’s really struggled. I don’t think he tried harder all of the sudden, or that those batters who went against him folded. Certainly there are learning curves for players, and “ah-ha” moments that may increase comfort, but the numbers just don’t bear out that this is something that turns on or off as a skill.

  9. Scot Lykins

    There are players that perform better than others under pressure. The players are not machine’s, they have emotions. There are people who can and cannot perform under pressure.

    Stats cannot measure the human element of the players.

    • jdx19

      Show me a ‘clutch’ player with a statistically significant amount of at-bats that cannot be explained by simple normal distribution happenstance.

      Just because you don’t understand it doesn’t make it incorrect.

    • jdx19

      And taking mathematics out of it… do you really think there is a big pressure differential between the following situations?

      1) 40,000 screaming fans, national TV, man on 2nd base
      2) 40,000 screaming fans, national TV, man on 1st base.

      One of those is traditional ‘clutch’ situation. The other is not a traditional clutch situation.

      So, a major league baseball player is nervous in #1 and not #2? Got it.

    • Steve Mancuso

      There are players who perform better under pressure. They are also the same players who perform better when they aren’t under pressure.

      Stats measure the outcome of the human elements being put to the test.

      If hitters empirically don’t hit better or worse over long periods of time when under pressure, what it is that the stats are missing? Hits, home runs, sacrifice flies, ground balls? Those can all be counted and compared to situations and between hitters. And those comparisons have been studied looking for clutch hitters. And researchers can’t find them.

      If clutch hitting is a thing that is so obvious, why can’t it be *easily* demonstrated by showing that certain hitters consistently perform better in big situations in comparison to lesser situations?

      • Scot Lykins

        If clutch hitting is a thing that is so obvious, why can’t it be *easily* demonstrated by showing that certain hitters consistently perform better in big situations in comparison to lesser situations?

        Because we can’t measure how a player feels. It is, in my opinion, not so much as performing better, as it is performing the same. More the ability to harness the nervousness and relax, rather than actually rising above natural ability.

      • Steve Mancuso

        They do. They perform the same over time. That’s what makes them professional baseball players. They can handle that stress of playing in those situations.

        People talk about Clutch as though it’s a great intangible thing. Well, how does it manifest? Do clutch players get more hits? more home runs? fewer strikeouts?

        The *fact* is that research on all the players, of all time, has failed to show these things. The research isn’t theoretical.

        When people started researching whether Clutch was a thing or not, they believed in it. It’s such a compelling narrative. They thought it would be easy to verify and quantify. Except when they actually looked for it, it wasn’t there.

      • CI3J

        Bill James said this: “How is it that a player who possesses the reflexes and the batting stroke and the knowledge and the experience to be a .262 hitter in other circumstances magically becomes a .300 hitter when the game is on the line? How does that happen? What is the process? What are the effects? Until we can answer those questions, I see little point in talking about clutch ability.”

        I do think there is a very simple physiological process that can easily explain clutch hitting: adrenalin. People respond to adrenalin rushes in different ways, and it has been documented that people can actually suppress the effects of adrenalin when put in a situation where it would be expected to occur (think someone disarming a bomb).

        Now, adrenalin suppression is something that has been theorized to be partially learned, and partially innate just like most human emotional responses. Essentially, someone who is continually exposed to high adrenalin situations can learn to channel adrenalin due to exposure. In essence, they build up “tolerance”. This is something that can be learned, and something we would expect all professional athletes to be able to do due to the very nature of their profession.

        However, an innate ability to not be so influenced by adrenalin, even when put into a “high adrenalin” situation that you haven’t experienced before, is something that is not universally shared by all people. Think if someone pointed a gun at you (I’m hoping most of you don’t have much experience with that). Would you “freak out”, hyperventilate, start to shake? Or would you narrow your eyes, calmly asses the situation as best you are able and try to figure out a positive outcome? Different people would respond to that situation in different ways. Think of how many times people have said “It all happened so fast!” of a high adrenalin situation, where another person would say “Everything seemed to move in slow motion.”

        This could be the “human element” of clutch hitting, as it were. However, if you want to discuss things statistically, I managed to dig up a pretty fascinating study on it. If anyone cares to discuss it, I can save it for another post as I think this one is long enough.

  10. Steve Schoenbaechler

    You sort of make a very good point here. Like when I was a kid, BA was most always used to show the best hitter. Then, OBP became that. Then, it was SLG. Then, OPS. Then, some people think it’s this new stat, that new stat, etc. Like here, is being “clutch” more important in the first inning or the last inning? If so, then wouldn’t it be more clutch to make that hit with two outs than no outs? If so, then we are talking about the clutch hitters being defined by those who drive in runs with 2 outs in the 9th inning. Like that happens to each hitter every game.

    This is why you simply can’t dig so deep with numbers. If you do, you lose sample size. At times, you have to say, “Enough is enough,” not deciding to throw numbers out but deciding that going deeper is simply too much. For also, numbers never tell the full story. For instance, with the Schumacher example above, he got a hit, but he didn’t drive in the run. Using solely that example, does that make Schumacher less clutch? Not at all. % of runners batted in? Well, one batter could have done that with one runner, making it 100%. So, shouldn’t there be a minimum limit on that? Then, like I said before, wouldn’t it be more clutch with 2 out instead of no out, in the 9th inning instead of the first inning? Some will say yes, some will say no. Who’s right? It’s all perspective.