Today In Poorly Formed Thoughts

I think one of the biggest myths that baseball fans hold dear to their hearts is that of the clutch hitter. It makes for a great story and lends well to legend to assume that certain players perform at their best in high leverage situations.

Unfortunately for gooey narratives, a player’s performance in one clutch situation has little predictive value on how he’ll perform in another clutch situation. To go even further, far too many players have similar numbers both “in the clutch” and in the regular at bats that it stretches the credibility of people who use the term “clutch” to describe anyone. Good players perform well when the game is on the line because they’re good players. It seems ridiculous to me to suggest that certain players save some sort of extra effort for high leverage situations whether than playing with that mindset all the time.

Even in those rare instances where there are noticeable difference between numbers in and out of clutch situations, it can normally be explained by looking at the type of player. For instance, a guy with a high batting average is more likely to get a memorable game winning hit than a player with a high on base percentage.

A couple of years ago, Tom Tango went to the trouble of polling the blogosphere on who fans would want at the plate with the game on the line. He took the results and compared them to the actual best batter on the team. The difference between the fans’ belief and actuality accounted for an average difference of 20 points in weighted on base average.

That says a lot about how we think of clutch performances. So much of it is based on our own perceptions without realizing how fallible our perceptions are. It’s not our fault, it’s our natural instinct as humans to garner opinions based on our observations. It’s just unfortunate that those observations are almost always either limited or biased.

Nonetheless, the myth of the clutch hitter still exists and was recently propagated by the likes of Mike Silva of NY Baseball Digest and New York Times columnist Paul Sullivan in a podcast. Sullivan was on the show to promote his book “Clutch: Why Some People Excel Under Pressure and Others Don’t.” The writer makes the following claims, as described by Silva:

There are five traits of a clutch performer: focus, discipline, adaptability, the ability to be fully in the present, and being driven–not thwarted–by fear and desire. Clutch doesn’t just mean hitting the game winning home run, but it could be a hit and run single that puts runners at the corners with none out in the first inning. That event could lead to the starting pitcher unraveling and blowing the game open early. He describes clutch performers as grinders. Individuals that don’t think of the glory that they will get, but rather are just focused on that one task and one moment. Perhaps that is why some of the most unlikely individuals become heroes when the stakes are highest during a short postseason series.

I think there’s an important distinction to be made that’s kind of missed by Sullivan. No one would ever suggest that clutch performances don’t happen. Players make timely hits or collect timely outs all of the time. It’s the way that the game is set up.

The difference is that if you’re going to label a player a “clutch performer,” it connotes that his being clutchy is repeatable. And, with all due respect to Pat Tabler’s numbers with the bases loaded, there isn’t a whole lot of evidence to suggest that the numbers from a large enough sample size of high leverage situations for any particular player is all that different from his career numbers that account for every situation. And even when a player’s clutch numbers are better than average in one year, they can be completely different the next.

Using FanGraphs definition of clutch, here are the leaders from 2010:

  • Michael Bourn 2.20
  • Placido Polanco 2.17
  • Carlos Lee 1.80
  • Jose Lopez 1.78
  • Delmon Young 1.74

In 2009, here’s how the players scored:

  • Michael Bourn 0.57
  • Placido Polanco 1.20
  • Carlos Lee 0.86
  • Jose Lopez -0.49
  • Delmon Young 0.48

In 2008:

  • Michael Bourn -0.49
  • Placido Polanco -0.30
  • Carlos Lee 1.37
  • Jose Lopez 0.69
  • Delmon Young 0.02

Can a baseball player be “clutch” one year and a “choker” the next? It seems more reasonable to me that over a smallish sample size success and failure can happen to both good and bad baseball players. It’s not a simple matter of being “clutch” or a “choker.”

On the flip side a “choker” is unable to accept responsibility and has a tendency to over think and be overconfident in those key situations. We have seen many individuals fall into this category throughout the history of sports, especially in this town.

Silva also uses Alex Rodriguez as an example of a “choker.”

Alex Rodriguez was the center of this debate for many years as he struggled in the postseason from 2005 to 2007 going 7 for 44. I will always remember how tense A-Rod was in a pregame interview on ESPN before the ’06 ALDS against Detroit. I knew he would struggle just from the poor body language he displayed. Alas, he would finish 1 for 14 in that series as the Tigers would shock the Yankees in four games. Things changed in 2009 as the post steroid “truth shall set me free” Rodriguez carried the Yankees to their 27th World Series title. After so many failures he finally was able to perform at his normal level during the heightened pressure of the playoffs. Many wondered what had changed.

This is exactly why the baseball season is 162 games long and players get 700 opportunities a season to prove their worth. Over the course of forty four at bats, anything can happen, even to a player of Alex Rodriguez’s superiority. It’s impossible to make his overall psychological tendencies in every at bat anything close to tangibly measurable.

Is there a single team in baseball that would rate a player based on 44 at bats? Then why should we as fans label someone clutch or choker based on the same amount of data?

Comments (21)

  1. While I like and agree with the premise that clutch hitting is relied upon and referred to far more than it should be, I think statheads often go too far when they claim that clutch hitting does not exist.

    Results are obviously skewed when evaluating small sample sizes (ie playoff sample sizes or clutch situation sample sizes), but that isn’t to say that over the course of a larger sample size there wouldn’t be a discrepancy between some players and others in high pressure situations.

  2. what about this…the “clutchiness” of the batter is opposed with equal “clutchiness” of the pitcher or defenders thus eliminating any “clutchy” advantage.

    • It’s so true. We tend to think only in certain terms that one player has to be clutch or a choker, but what about the other player in the clutch situation? I can’t imagine that if I was a manager I’d put the weight of numbers in the clutch over a player’s overall numbers when making a decision.

  3. Not to mention the fact that late game situations usually feature closers – closers often use fastballs more than other pitchers – some guys hit fastballs way better than others. John McDonald comes to mind as a guy who has performed well with the “game on the line”. But he’s a fastball hitter.

    Sorry that made little sense, but I guess my point is that if Mariano Rivera is pitching, I’d rather the guy who hits the fastball well versus the guy with the good leverage stats.

  4. joe carter’s career ops was 40 pts higher with risp than with nobody on.

    If I was to think of a clutch jay hitter it would be roberto alomar.

    his ops was 66pts higher. he ops’d 1.006 with the bases loaded.

    I think it ‘s pretty clear there are clutch players.

  5. tabler ops’d a whopping 144 pts higher with risp and 1.198 with bases loaded.

    the fact is some players play better when the pressure is on and others don’t

    • Nope. The fact is that random things happen and skew results in small sample sizes. It’s easy to pick out individual situations to make an argument, but you’re ignoring so very much in doing so.

  6. now let’s think of a jay choker, how bout alex rios?

    ops’d 30 pts lower with risp bases loaded .618 ops

  7. Carter was clutch that one time in 93.

  8. yes dustin .010 is statistically significant. now who’s trolling? you’re right, carter wasn’t very clutch. that home run he hit to win the world series was pretty uneventful right? you must really enjoy losing.

  9. carter and alomar had 1000′s of ab’s in said situations. that’s not small sample sizes.

    • You’re still cherry picking situations. And you can’t prove a consistency from year to year. Let’s try to find something we can agree on and work from there. Do you have any problems with how FanGraphs measures “clutch?” If it’s a skill, shouldn’t it be visibly consistent from year to year? Clutch is never consistent.

  10. If you are discussing the importance of sample sizes, then it wouldn’t necessarily be consistent year to year at all. That is, the number of “clutch” situations may provide for a fairly small sample over the course of a year, and results will vary accordingly.

    Again, I agree with the overall approach. I just think you’re dead wrong to dismiss it entirely. I feel pretty much the same way about character by the way – too many people read too much into it, but there is still some value to the idea that you are dismissing entirely because the stats can’t appropriately quantify the impact.

  11. how is 1000′s of ab’s cherry picking? you warn of the pitfalls of using small sample sizes then at the same time ask me for year to year results which would be a smaller sample size thus making those results more inconsistent! the only thing you can do is aggregate the data and see if there is a long term bias.

    clutch doesn’t have to be consistent. the weather is cold right now. does that mean there is no global warming? shouldn’t it be consistently hot every spring to prove global warming? that’s why you look at long term trends to see if there is a pattern.

    • @ Grouchy. I tell you what. If you can find someone else to take your side on what you’re saying, I’ll be glad to continue the argument. Wow. No one is claiming that weather patterns have the same properties as skills.

  12. Here’s what I don’t get about “clutchiness.” If players were so good in these pressure situations, why weren’t they as good in situations that weren’t as pressured. Laziness? Lack of focus? Didn’t care? I don’t buy it. Small sample size and random variation makes more sense.

  13. Overachieving in the “clutch” is mostly bullshit. Why the hell won’t that player play like that all the time then? However, underachieving in the “clutch” can be explained a bit. My example is Mike Napoli. In high leverage situations, he has a way higher FB% and a way lower LD% & GB% than he does in lower leverage situations.

  14. I think the definition of ‘clutch’ has gotten way too broad. RISP in the bottom of the second inning doesn’t seem clutch, at least not in the way RISP in the bottom of the ninth trailing by 1 run is. Would ‘clutch’ players be better the more ‘clutch’ the situation is? Would their numbers be higher with 2 outs in the bottom of the ninth than with no outs? When their team is down a run instead of up by five? Do we count entire games against division rivals battling for a playoff spot clutch? For that matter, shouldn’t a ‘clutch’ hitter’s career post-season numbers be substantially better than his regular season numbers?

    Clutch is way too slippery a concept to be borne out by a simple analysis like OPS with RISP. From what I understand about high-caliber athletes, and that’s admittedly not much, the goal is to block out or ignore all mental/emotional pressure, not respond to it. The best athletes perform well under pressure because they are able to ignore that pressure, and treat a clutch at-bat like any other at-bat.

  15. Actually the concept of “leverage” is designed to account for these differences in the level of importance in game situations.

    I’ll play a bit of devil’s advocate here. As someone well versed in the field of stats, I have to say that the way “leverage” and “clutch” is calculated leaves something to be desired. I am not yet convinced it does not exist. It is like the god particle, maybe it is there and we just haven’t found it yet.

Leave a Reply

Your email address will not be published. Required fields are marked *