Clutch vs. choke - do these numbers mean anything?

[moonslav59]

CERA is terrible.

 

 

So, it's real, but just terrible.

[User Name]

I mean it's real because it's an actual stat with an actual formula used for its calculation. But it's terrible, because it has significant background noise. Too many factors for it to be effectively objective.

[moonslav59]

I mean it's real because it's an actual stat with an actual formula used for its calculation. But it's terrible, because it has significant background noise. Too many factors for it to be effectively objective.

 

It has a very limited use: it's only good when comparing catchers from the same team with the same pitchers only. The sample sizes have to be significant, so often there are very few samples to compare.

 

Other than that, the variables are not much different from other stats: strength of opponents vary, park factors, defensive alignments behind the pitchers and more vary, but that's true of BA, ERA, OPS, etc...

 

It's also best to use OPS against- not just CERA.

 

When one catcher continuously gets blown away by just about every pitcher by pitcher comp, it's hard to ignore.

[iortiz]

His Tuesday sample size is large, too.

 

It has nothing to do with pressure situations.

[moonslav59]

It has nothing to do with pressure situations.

 

Scattered small sample sizes are similar.

[iortiz]

Scattered small sample sizes are similar.

 

Nope, career Tuesdays are not a good sample to represent high leverage situations. POs are.

 

While POs are not large samples by definition, POs represent the highest moments of pressure in baseball. In Kershaw’s case (as others) the sample is fair in order to argument that he simply is not very good as he is in regular season.

[Bellhorn04]

CERA is terrible.

 

Clutch is not a repeatable skill.

 

But repeatable skill is not well defined. It's a term I've never heard used except in this context.

[moonslav59]

Nope, career Tuesdays are not a good sample to represent high leverage situations. POs are.

 

While POs are not large samples by definition, POs represent the highest moments of pressure in baseball. In Kershaw’s case (as others) the sample is fair in order to argument that he simply is not very good as he is in regular season.

 

Random samples would produce the exact amount of Kershaws as there are. How do you explain that away without ever addressing that clear fact?

[Bellhorn04]

Random samples would produce the exact amount of Kershaws as there are. How do you explain that away without ever addressing that clear fact?

 

Honestly, I don't even know what that first sentence means.

[moonslav59]

Honestly, I don't even know what that first sentence means.

 

If a computer generated random samples of playoff results based on regular season numbers and specific sample sizes, it would generate a sample nearly identical to Kershaw's playoff line.

 

How can anyone not think that maybe Kershaw's line is just a result of being that one guy that came out with bad numbers. It may have nothing to do with pressure or clutch.

 

Players slump and streak all the time for reasons unknown to anyone, even themselves. Wouldn't one think players might slump during the playoffs for reasons having nothing to do with the fact that there is extra pressure on them and everyone in those games? Why do we, all of a sudden, pretend to know the specific reason Kershaw slumped at these times but not the others?

 

If Kershaw does great, this October, will you guys say he gained the skill of handling pressure better?

 

If he does poorly, I'm sure we'll here the, "See, I told you so's."

 

Over Kershaw's career, he has allowed a .582 OPS against, but when it counts, during the regular season, sometimes in big games that determine making the playoffs or not, he has these numbers in high pressure situations:

 

.547 Late & Close

.502 RISP, 2 outs

.582 RISP

.585 Men on Base

.599 High Leverage

 

Nothing here indicates he comes up short, when it counts, but because of a scattered 189 IP sample size over a 14 year career where his WHIP is 1.072 as compared to a 1.002 regular season one, we are supposed to believe it HAS TO BE BECAUSE HE's A CHOKE! There can be no other explanation. It's a fact, because the numbers speak for themselves. He couldn't have just had the bad luck of slumping too many times in October, instead of September or May.

 

Look, I know I can't prove clutch & choke are not real, and it does make some sense to think some players handle pressure better or worse than others, but it can't be proved the reason Kershaw's ERA is 4.19 in the playoffs is because he chokes or is NOT clutch. It just can't.

 

In 8 of his 20 playoff series, he had an ERA of 3.00 or less, including 2 with a 0.00 ERA, which included an 8 inning WC shutdown. You can't get anymore clutch than a shutdown in an elimination game, but I guess that was just luck- not skill.

 

My point is, if a computer generated the same amount of sample sizes as there are in the history of MLB playoffs, there would be some samples that jump out as being way lower and higher than the norm. It's just the way randomness works. Players often do real good for some stretches and real bad in others. There may or may not be any rhyme or reason for it. Why should the playoff samples, often spread over many seasons be any different?

 

In fact, they are no different from computer generated outcomes.

 

Coincidence or just randomness playing itself out?

[Bellhorn04]

Look, I know I can't prove clutch & choke are not real, and it does make some sense to think some players handle pressure better or worse than others, but it can't be proved the reason Kershaw's ERA is 4.19 in the playoffs is because he chokes or is NOT clutch. It just can't.

 

Yep, can't be proved, can't be disproved. I think we're in agreement on that.

 

But all the stuff about computer generated random samples just seems like gibberish.

[5GoldGlovesOF,75]

But repeatable skill is not well defined. It's a term I've never heard used except in this context.

 

Right; how many times does a batter need to come through... in pressure situations... to be considered "clutch"?

 

Compare Big Papi and ARod. Each had over 1,000 regular season plate appearances in the big leagues. Their stats are very similar in what most observers would consider high leverage situations: Ortiz batted a little better with 2-outs, RISP, .276/.951 to .265/.849; ARod had a miniscule edge in Late & Close, .265 to .256 (same OPS). In tie games, the two big boppers had identical batting averages of .292...

 

I couldn't find postseason stats for these categories -- but does it really matter (maybe that's the point)? If you polled 1,000 diehards who watched their careers -- including Red Sox and Yankee fans -- and asked which guy they'd want up at bat in a "clutch" moment in the playoffs, how many would pick Alex Rodriguez over David Ortiz?

Seriously... any?

[Bellhorn04]

Yes, the term "repeatable skill" is totally made-up and arguably useless.

[notin]

Yes, the term "repeatable skill" is totally made-up and arguably useless.

 

The entire notion of using post-season stats to determine clutch is also made up and useless. We've all seen plenty of players get post-season at-bats in blowout games.

 

Really the notion of "clutch" in itself is pretty fabricated. A lot of times, it comes down to simply how you remember a player. I've seen people say Yaz wasn't cltuch because he popped out to end the 1978 season with the tying run on third. These people saw 39yo Yaz in that game, and, like me, clearly missed his monstrous ending to the 1967 season...

[Bellhorn04]

The entire notion of using post-season stats to determine clutch is also made up and useless. We've all seen plenty of players get post-season at-bats in blowout games.

 

Really the notion of "clutch" in itself is pretty fabricated. A lot of times, it comes down to simply how you remember a player. I've seen people say Yaz wasn't cltuch because he popped out to end the 1978 season with the tying run on third. These people saw 39yo Yaz in that game, and, like me, clearly missed his monstrous ending to the 1967 season...

 

Nice whataboutism.

 

People who say that about Yaz are what you call...stupid.

 

Yaz also made the last out in the 1975 WS, BTW.

Edited August 17, 2021 by Bellhorn04

[moonslav59]

Yep, can't be proved, can't be disproved. I think we're in agreement on that.

 

But all the stuff about computer generated random samples just seems like gibberish.

 

It doesn't make any sense to go that computer generated models based on random numbers based on regular season numbers comes out to look nearly identical to what real life sample sizes variances show means nothing?

[5GoldGlovesOF,75]

I've seen people say Yaz wasn't cltuch because he popped out to end the 1978 season with the tying run on third.

 

The glory of Captain Carl. Only his most devoted fans remember in Game 163 that year when he pulled the best pitcher in baseball for a home run to give the Sox an early lead, and then singled off Gossage to drive in another run in the bottom of the 8th. Then he scored Boston's last run of the season from 2nd to make it 5-4. Some would say each of those feats show times he came through -- if you consider it important to score first in a do-or-die game or produce runs with your team trailing late...

[mvp 78]

The glory of Captain Carl. Only his most devoted fans remember in Game 163 that year when he pulled the best pitcher in baseball for a home run to give the Sox an early lead, and then singled off Gossage to drive in another run in the bottom of the 8th. Then he scored Boston's last run of the season from 2nd to make it 5-4. Some would say each of those feats show times he came through -- if you consider it important to score first in a do-or-die game or produce runs with your team trailing late...

 

2 for 5, 1 HR, 2 R, 2 RBI, 1 K

 

Seem like a good game in a do or die situation.

 

Maybe we'd have to go back and check is barrel rate?

[Bellhorn04]

It doesn't make any sense to go that computer generated models based on random numbers based on regular season numbers comes out to look nearly identical to what real life sample sizes variances show means nothing?

 

Well, I'd like to see this applied to Kershaw. As I said, his worst monthly ERA is 2.84, so I'm not sure how you're going to generate a sample that has a 4.19.

[moonslav59]

Well, I'd like to see this applied to Kershaw. As I said, his worst monthly ERA is 2.84, so I'm not sure how you're going to generate a sample that has a 4.19.

 

He, himself had a season of 4.26 and 3.39.

 

He's had many bad starts. You don't think randomly choosing any 30 starts (the amount he's had in the P.O.s might generate one with an ERA near 4.19 and maybe one near 1.19?

 

[Bellhorn04]

He, himself had a season of 4.26 and 3.39.

 

He's had many bad starts. You don't think randomly choosing any 30 starts (the amount he's had in the P.O.s might generate one with an ERA near 4.19 and maybe one near 1.19?

 

 

I think it would be a tiny percentage. If you can come up with a tiny percentage of samples that fit, what does it show? That there's a tiny chance the postseason numbers can be attributed to randomness?

[moonslav59]

I think it would be a tiny percentage. If you can come up with a tiny percentage of samples that fit, what does it show? That there's a tiny chance the postseason numbers can be attributed to randomness?

 

The percentage of actual players who have wildly different playoff ERAs than regular season ones is nearly identical to the computer generated samples.

 

If only a tiny percentage of actual pitchers have massive disparities, like Kershaw, then so does the randomly generated one. That's the point. They mirror each other almost exactly the same.

 

The computer comes up with maybe 3 guys who have ERAs way higher than their norm, and in reality, there are 3 guys who fit the model. That's, at least, what I understand the studies have shown.

[iortiz]

Random samples would produce the exact amount of Kershaws as there are. How do you explain that away without ever addressing that clear fact?

 

Random samples? clear fact? lol

 

I do not even know what you are talking about.

[iortiz]

Yes, the term "repeatable skill" is totally made-up and arguably useless.

 

Well Kershaw time after time in POs (all-in-all) hasn't been that good. OTOH He's been very good in Regular season.

 

If that is not repeatable then I do not know what it is.

[User Name]

But repeatable skill is not well defined. It's a term I've never heard used except in this context.

 

Because this is the correct context for its usage.

[User Name]

Well Kershaw time after time in POs (all-in-all) hasn't been that good. OTOH He's been very good in Regular season.

 

If that is not repeatable then I do not know what it is.

 

You're arguing two differente points.

[iortiz]

You're arguing two differente points.

 

yup, in order to contrast different environments.

[Bellhorn04]

Because this is the correct context for its usage.

 

But what does it mean?

[moonslav59]

 

I do not even know what you are talking about.

 

Obviously.

[moonslav59]

yup, in order to contrast different environments.

 

How do you prove it's the "environment" that caused the difference?

 

Nobody knows why players slump and get hot during the season, but suddenly, when they do in the playoffs, we know for sure why.

 

That really makes sense to you?