My own limited study of "clutch"

[Brutus]

I have been toying around with a lot of baseball data the last few weeks and months. The last few days I got bored and did a limited sample of "clutch hitting" and trying to determine whether or not it exists.

First, I expected to find it does exist, at least in some capacity. The hardest part about clutch hitting is that it's a subjective definition and it means something different to every person. But if it doesn't exist, you wouldn't expect to find much of a fluctuation in batting average, OPS, ISO power, etc.

I know many people have done different tests on this very subject, and there have been mixed results. And I also know these findings I've got are neither conclusive nor ground-breaking.

But here are the results:

(I have not run any sort of regression analysis, correlation testing or anything of that nature... just wanted to find basic splits)

Over 25 years worth of data, dating back to 1983:

First, I examined Bases Occupied data against no runners on base. I wanted to find stats with runners in scoring position.

None On: .258 BA, .321 OBP, .405 SLG, 726 OPS, .147 isoP, 34.8 AB/ HR
None On or 1B occupied: .263 BA, .322 OBP, .411 SLG, 733 OPS, .148 isoP and 34.6 AB/ HR

Now, the splits with Runners in Scoring Position (any combination of 2B, 3B, 2nd and 3rd or Bases Loaded):

RISP: .265 BA, .346 OBP, .408 SLG, 754 OPS, .144 isoP, 37.5 AB/HR

My theory on the spiked On-Base Percentage is that often times with a base open and runners at 2nd/3rd or both, there are a lot of intentional base-on-balls. I didn't account for that before I cumulated these numbers, but I would imagine if I removed the IBB, these would be right in line with the other stats. The slugging, and isolated power numbers are very similar, meaning the rise in OPS is mainly from the 20 points in OBP (presumably from intentional walks).

So while I found that RISP data didn't give any indication of clutch hitting, the number of outs in an inning and the innings in a game did show a fair amount of difference.

Take RISP and isolate that split to just two outs. Here's what I found:

RISP 2outs: .242 BA, .349 OBP, .379 SLG, 728 OPS, .136 isoP, 39.4 AB/HR

As you can see, the batting average took a huge hit as did slugging, isoP and No. of At-Bats per home runs. The OBP again was much higher, but once again, I believe this to be a product of intentional walks.

Now the splits by Out:

0 outs: .271 / .322 / .424 / .746 / .152 iso / 33.7 AB per HR
1 outs: .268 / .331 / .414 / .745 / .146 iso / 35.8 AB per HR
2 outs: .250 / .332 / .391 / .724 / .141 iso / 36.5 AB per HR

As we see with RISP, production with two outs is tremendously deflated. The batting average dropped 21 points from no outs and while On-Base Percentage actually goes up a little, slugging (and accordingly, isolated power) both drop a lot. Home Runs also become a little more scarce.

Lastly, splits by Inning:

Innings 1-3: .264 / .329 / .411 / .740 / .147 iso / 35.7 AB per HR
Innings 4-6: .269 / .331 / .425 / .756 / .155 iso / 32.9 AB per HR
Innings 7-9: .256 / .324 / .394 / .718 / .138 iso / 37.3 AB per HR
Late & Close: .254 / .326 / .385 / .711 / .130 iso / 39.2 AB per HR

In this section, we see that the Averages between the first three innings and second three are fairly similar, though there's a slight spike in batting average and power categories. This makes sense as starting pitchers' pitch counts go up and start to tire, they're more prone to giving up hits and especially home runs. It makes sense then, that it's going to be harder to hit in the late innings because batters are facing fresher pitching (usually).

That's where we have to determine if the drop in late innings have more to do with getting to the bullpen and livelier fastballs or if it's because of "clutch" situations.

In any event, while we don't know the answer exactly, we do know that batting average in late & close situations dips all the way down to .254, and the Slugging and ISO power numbers go way down, resulting in a 711 OPS. At-Bats per home run goes way up to 39 for every home run.

A lot is still to be found, I imagine, on this subject. But while I can't say conclusively clutch exists (and even after this, I still believe it does), it's clear that statistically, it's much harder to hit by out/inning situations when you get later into either. So the guys that do have any history of excelling in those particular situations should, if nothing else, be considered strong performers in "clutch" situations if not outright called "clutch hitters."

Disclaimer: this thread was not meant to be about Adam Dunn, though I'm sure some people will automatically bring up his name given his controversial nature to hit (or not hit) in "clutch" situations.

[reds2221]

very nice research, you really put alot of work into this. I did notice however that as far as outs go, the OBP goes up as the outs do.

[Brutus]

very nice research, you really put alot of work into this. I did notice however that as far as outs go, the OBP goes up as the outs do.

That was something I noticed too. The only guess I have on why that might be has to do with not only intentional walks, but non-intentional, intentional walks where pitchers have two outs and are more likely to pitch around certain hitters. I did see a spike in the number of intentional walks, so I imagine that a spike in overall base on balls would corroborate this theory at least a little.

[757690]

Excellent research. This backs up a theory that I have been preaching for years, that it is harder to get a hit in clutch situations since the opposing team knows it's clutch and works harder to get you out.
They bring in the best pitcher match up wise to face you, they are more alert on defense, they use their best "out" pitches against you, and are more focused on getting you out in general.
Your research provides some evidence that this is true. While no statistical research is 100% conclusive, this is pretty strong stuff.
I think this changes what it means to be a clutch hitter. I think that if a hitter can out perform the production you have shown to be the MLB average for that situation, then he is clutch. It has nothing to do with how well he hits in other situations, it is how well he does compared to the rest of the league.

Thanks, great stuff!

[Lockdwn11]

Good stuff

[Brutus]

I think this changes what it means to be a clutch hitter. I think that if a hitter can out perform the production you have shown to be the MLB average for that situation, then he is clutch. It has nothing to do with how well he hits in other situations, it is how well he does compared to the rest of the league.

Thanks, great stuff!

My idea on measuring this would first be to take a batting average in whatever is defined as a "clutch" situation, and divide it by the league average for that exact split. Then you get a league normalized factor.

Then, I would subtract that to the player's batting average (minus the clutch situations) adjusted to overall league batting for that environment.

What you have, then, is a league-adjusted "clutch" batting average that defines not only the player against his league, but also measures whether a player exceeds his own "normal" production when clutch situations are present. This eliminates the argument that the best players in clutch situations are also the best players in normal situations... or at least, it takes the possibility of that theory into account.

[reds2221]

That was something I noticed too. The only guess I have on why that might be has to do with not only intentional walks, but non-intentional, intentional walks where pitchers have two outs and are more likely to pitch around certain hitters. I did see a spike in the number of intentional walks, so I imagine that a spike in overall base on balls would corroborate this theory at least a little.

thats what I assumed the reason for that was too

[Orodle]

I think its stupid to say clutch doesnt exist. Its like pretending humans arnt playing the game. I believe everyone can agree that certain people can handle certain situations better than others both physically and mentally.

[757690]

My idea on measuring this would first be to take a batting average in whatever is defined as a "clutch" situation, and divide it by the league average for that exact split. Then you get a league normalized factor.

Then, I would subtract that to the player's batting average (minus the clutch situations) adjusted to overall league batting for that environment.

What you have, then, is a league-adjusted "clutch" batting average that defines not only the player against his league, but also measures whether a player exceeds his own "normal" production when clutch situations are present. This eliminates the argument that the best players in clutch situations are also the best players in normal situations... or at least, it takes the possibility of that theory into account.

That sounds like a great way to do it. Thanks