Tuesday, February 05, 2008

The clutch hitting bet, part III

A few weeks ago, I offered to bet against anyone who believed there was such a thing as a clutch hitters who could be identified in advance. I got a total of about three bets. That's out of, probably, a few hundred people who read about my offer.

My conclusions, after reading some of the comments on other blogs that mentioned the bet:

1. People may say they believe some players are clutch hitters, but they don't really believe it. If they did, they'd bet.

A common excuse that some people give is that they don't bet, as a matter of principle. This is fine if it comes from one person, but when two hundred people all refuse to bet, it's gotta be more than principle.

Similarly are the people who argued that, if I believed the odds were close to even, I should offer better odds than 2:3. That's a silly excuse. If I had decided that Juan Pierre had a 50/50 chance of outhomering Ryan Howard, and offered the same 2:3 bet, nobody would have time to haggle over the odds – everyone would be too busy rushing to get their bets in.

2. People don't even know what they mean when they talk about clutch hitting. There were a few commenters (on other blogs) who said that, yes, *of course* there are clutch hitters, but you don't measure their clutchness by comparing their regular output to their clutch output.

But if clutch doesn't mean being BETTER in the clutch, what does it mean? As Tango and/or MGL have pointed out many times, if you just mean that they're good in the clutch, doesn't that just mean they're good players in general? I mean, of course you'd rather have David Ortiz up in an important situation than Neifi Perez. But you'd rather have David Ortiz up in ANY situation than Neifi Perez. So what's the point of calling him clutch?

Perhaps many clutch advocates have only a fuzzy idea of what something means, but get confused when you try to pin them down on it? It seems like they just never thought farther than their initial, happy fuzzy reaction.

3. As mgl said in comments on several blogs: if you are unable to pick any player, or any combination of players, whom you believe have even a 60% chance of being clutch (when the average is 50%), then you are pretty much admitting that clutch hitting talent isn't a very strong factor in baseball. To
quote mgl,

"[If you don't think the odds are good enough,] you are essentially agreeing that a clutch player is virtually indistinguishable from a choke player before the fact! If you can't beat 55-50 over en entire season with a bunch of so-called clutch players verus choke players, you have no right to talk about who is clutch and who is not!"

Put another way: by complaining about the odds, you are admitting that you don't know who the clutch players are, or that they are not significantly better than other players. In that case, if you still talk about clutch hitters as if they are a significant force in baseball – if, for instance, you continue to refer to David Ortiz as a clutch hitter as if that skill forms a measurable part of his value – you are either bullshitting, or have compartmentalized your brain into believing two contradictory things at once.

4. Speaking of believing two contradictory things at once: there is ample precedent for that sort of thing. People will tell you how great heaven is, and how it's everlasting bliss and contentment, and how they will be reunited with their deceased relatives that they miss so much, and that heaven is the ultimate reward. But they act as if death were the worst thing that could possibly happen to them.

It makes people feel good to think that clutch exists, and so they'll keep talking about it, and psuedo-believing in it, and ducking and weaving when we try to pin them down.

If you are a clutch-hitting advocate who believes this is unfair, you may be right. Maybe it is unfair, applied to you. To prove it, show us your belief is rational and open to new evidence. The best way is to accept my bet. If you say you have moral scruples against betting real money – you never even buy raffle tickets, and you believe even betting for charity is immoral – well, I'm reluctant to believe you. But try this instead.

Write down your best prediction about clutch hitting in 2008. That is, decide what you are most sure of about when it comes to what will happen in the clutch this year, something you believe that would not be true if clutch were close to random. Write it down, and stick it on your fridge.

And assure yourself that if your prediction does NOT come true, you will publicly re-evaluate your belief in clutch hitting talent. If you can't tell yourself that – if your belief in clutch hitting is immune to your absolute BEST prediction in it failing to come true – then either you believe in a very, very weak form of clutch hitting, or your belief in it is faith-based and unfalsifiable.

Labels: ,


At Wednesday, February 06, 2008 12:47:00 PM, Blogger Unknown said...

I agree. You got two takers? I really am shocked at the low turnout for this.

Oh well. It will still be interesting to see how the two sets turn out.

At Wednesday, February 06, 2008 2:24:00 PM, Blogger Phil Birnbaum said...

Well, whichever way they turn out, the experiment has pretty much proven that people DON'T really believe in clutch hitting, no matter what they say.

Except for the two that bet.

At Thursday, February 07, 2008 11:21:00 AM, Anonymous Anonymous said...

Let me explain why I believe in clutch hitting, but didn't take this bet. In my opinion, the reason clutch hitting is so hard to measure, and the problem with many of the studies that have tried to discover it, is that it is very difficult to figure out what counts as a clutch situation in a 162 game season. There are very few games that are must-win situations, with the exception of games against division rivals in close races late in the season.

This means that the best strategy for a player is to conserve mental energy and treat each at-bat the same, except for maybe a few games. The problem is, it's impossible to predict those games in advance. For example, I think that Ortiz is more clutch than A-Rod. But I wouldn't predict Ortiz to be locking himself in for every late-inning at-bat with runners on base in April games against the Royals, so his overall season "clutch" stats might not reflect his clutchness. I could try to limit the range to games in August and September, but then I run into the opposite problem. What if the Red Sox are fifteen games out of first place by the All-Star break? Surely, you wouldn't call too many at-bats "clutch" in a lost season.

I do think the clutch effect is measurable in postseason games, given a sufficiently large sample size of at-bats. In the post-season, every game is very important, and I would expect players to lock themselves in and really try to bring their focus to the next level, or alternatively to fold under the pressure. I would also expect this effect to apply to every postseason at-bat, because the added pressure is there throughout the game. (Note that in general you would expect a slight decrease in postseason performance because the pitching is likely to be better as well, so any increase in performance is particularly impressive.)

To test this, I came up with four players who I think of as historically clutch: Babe Ruth, Reggie Jackson, Kirk Gibson, and David Ortiz. Each of these players also has enough at-bats in the postseason to make for a reasonably meaningful comparison. I checked their postseason versus regular season numbers in four categories: BA, OBP, SLG, and At-bats per home run.

Their career postseason numbers are (BA, OBP, SLG, ABs per HR):

Regular season: .342 .474 .690 11.8
Playoffs: .326 .467 .744 8.6

Regular season: .262 .356 .490 17.5
Playoffs: .278 .358 .527 15.6

Regular season: .268 .352 .463 22.7
Playoffs: .282 .380 .577 11.1

Regular season: .289 .384 .559 15.8
Playoffs: .317 .418 .587 17.2

Each player improves significantly when his team's season is on the line, most notably Gibson. This would seem to indicate that these players were able to elevate their games in the most important situations.

Compare to A-Rod's splits:

Regular Season: .306 .389 .578 14.2
Playoffs: .279 .361 .483 21

This obviously isn't scientific, but hopefully it helps explain why I think the data do support a belief in clutch hitting, albeit one that is difficult to establish over the course of a regular season when clutch situations are hard to adequately predict.

At Thursday, February 07, 2008 12:29:00 PM, Blogger Unknown said...

So, THANE, are you saying that if Boston and NY both qualify for the postseason this year, you would bet that Ortiz is clutch and ARod is not?

If you were, I'd imagine you could find a taker to oppose you.

At Thursday, February 07, 2008 12:52:00 PM, Blogger Phil Birnbaum said...

Thane, thanks for the comments.

But isn't there a bit of circular reasoning there? The reason you think of those four players as clutch is because they DID perform well in the clutch. That actually isn't evidence of clutch hitting, just evidence that your memories of their heroics are accurate.

In any case, if you want to pick any number of "clutch" players and any number of "unclutch" players, and make "playoff game" the clutch situation, I'll give you the 2:3 odds. I recommend you choose lots of players so that you have enough on each side with only 8 teams in the post-season.

At Saturday, February 09, 2008 4:10:00 AM, Anonymous Anonymous said...

You ask, "But if clutch doesn't mean being BETTER in the clutch, what does it mean?"

Clutch hitting means being BETTER (than others) in the clutch

You wish to use the definition that clutch hitting means being BETTER (than yourself)

At Saturday, February 09, 2008 1:33:00 PM, Anonymous Anonymous said...

Phil, I might still take the bet, using the regular season/postseason split. (Or, more likely, a pre-All Star break versus postseason split).

I'll have to do some thinking and see if I can come up with a list of clutch players that includes enough to be sure of a representative sample in the postseason.

I would prefer to use OPS, though, rather than BA, because I'm a firm believer that OPS measures patting performance significantly better than BA.

Also, on the circular reasoning point, obviously that's a problem. But if you accept my premise that there are likely to be few, if any, true clutch situations in the regular season, then it makes a lot more sense that sabermetricians haven't been able to prove the existence of clutchness. You reduce the available data so much.

At Tuesday, February 12, 2008 8:17:00 PM, Anonymous Anonymous said...

Tom G.,

If being better in the clutch than OTHERS is the measure for determining who is clutch and who is not, aren't we really just looking at who are the better players, which is essentially what statistically oriented people have been saying for years?

What if a player has a career batting line of .330/.400/.500, but hits .260/.320/.400 in so-called clutch situations, but another player with a career batting line of .240/.300/.360 hits in clutch situations .255/.315/.395. Isn't the 2nd player a better clutch performer than the 1st despite having worse numbers because the first players numbers declined in the clutch while the second players numbers improved?

If I understand you, you're basically agreeing with most sabermetricians in that the better the hitter, the better he'll do in the clutch.


Post a Comment

<< Home