Concepts some posters could benefit from learning

[Fencer]

it seems this test would be perfect for the 7/8 hits against Minnesota example, did you run that?

Much lower. Also substantively meaningless, because of the way you cherry-picked those two games.

For example, given that Yaz was in a couple of Game 7s, a one-game playoff, and so on, those were NOT the only do-or-die games he played in.

For example, taking just the five games I just mentioned, his aggregate batting was 11-20. Or if you look just at postseason elimination games, it's 13-26. That latter is not statistically significant when tested against his overall postseason batting average, but (and here we finally have support for your intuition of his clutchness in do-or-die games) probably is when tested against his regular season averages. (E.g, 2.5% or so if you use a BA of .300 or so, which seems like a reasonable estimate given which seasons those games followed.)

 

[RayClay]

Much lower. Also substantively meaningless, because of the way you cherry-picked those two games.

For example, given that Yaz was in a couple of Game 7s, a one-game playoff, and so on, those were NOT the only do-or-die games he played in.

For example, taking just the five games I just mentioned, his aggregate batting was 11-20. Or if you look just at postseason elimination games, it's 13-26. That latter is not statistically significant when tested against his overall postseason batting average, but (and here we finally have support for your intuition of his clutchness in do-or-die games) probably is when tested against his regular season averages. (E.g, 2.5% or so if you use a BA of .300 or so, which seems like a reasonable estimate given which seasons those games followed.)

Not at all. Do I cherry pick the only two games with any meaning in my Red Sox lifetime? That's like saying players choose when they would like to participate in crucial games.

I thought his performance was extraordinary in a sport where 3 out of 10 is a good performance.

I then checked to see if he also performed better than expected. I didn't know his playoff averages for sure and i don't doubt we can pick other circumstances that wouldn't fit. I don't think his playoff averages alone are extraordinarily significant.

The poster said it has been statistically proven there is no such thing as clutch. I think these situations are indicative and clutchness could be inferred based on a common measure such as batting average.

All I really wanted to test was the probability of getting 7/8 hits when the mean was 2.61/8 (for a .326 hitter).

I'm a social studies guy taking a required course in which we do not do the math (thank God).

How out of the ordinary is it is my only question?

 

[RayClay]

We can't possibly use this binomial distribution to prove or disprove clutchness right?

This test assumes the probability of the event is .285, but the argument of clutchness is that this probability rises in pressure situations. If we use this on the basis that a player cannot become better in pressure situations, then this only tells us that at least 24 successes out of 65 tries has a probability of .088, correct?


Using the binomial distribution in this discussion seems futile because it is operating under the pretense that "clutchness" does not exist.


To look at proof of clutchness or not, we must study all players' averages in "clutch" situations vs. all other situations. But this will only tell us that most players did not show much improvement over their norm in the clutch. The argument that CERTAIN players posessed this potential human quality can't really be proven or disproven based on their performance numbers. I'd guess it would be more a study of the human brain, affect of extra adrenaline, etc. etc...

Am I wrong here?

Well, a batting average is the mean of the hitters performance during the season.

Therefore, comparing his batting average in any situation to that gives you tons of tests which compare actual results to expected results (season or career batting average).

I know what they are, but i don't know how to run them.

It doesn't matter how other players perform unless you want tomake that comparison.

 

[emoney_33]

Well, a batting average is the mean of the hitters performance during the season.

Therefore, comparing his batting average in any situation to that gives you tons of tests which compare actual results to expected results (season or career batting average).

I know what they are, but i don't know how to run them.

It doesn't matter how other players perform unless you want tomake that comparison.

I understand that his results in the playoffs and world series were better than his "expected" results based on his career batting average. But this discussion is about where these "unexpected" results come from. How much luck is involved, how much "clutch ability" is involved, etc... We can compare numbers all we want but they will never prove one way or the other of WHY the numbers differ. The numbers tell us that in fact he hit better in the postseason, they don't tell us why, nor do they even suggest a reason for why. We can conclude that he WAS clutch, whether that was due to luck or some very rare ability to gain ability in higher-pressure situations, is not something we can conclude.

The clutch debate goes on because of the inability to prove it one way or the other, both sides have valid arguments and facts to back up the opinion of clutchness.

 

[emoney_33]

Not at all. Do I cherry pick the only two games with any meaning in my Red Sox lifetime? That's like saying players choose when they would like to participate in crucial games.

I thought his performance was extraordinary in a sport where 3 out of 10 is a good performance.

I then checked to see if he also performed better than expected. I didn't know his playoff averages for sure and i don't doubt we can pick other circumstances that wouldn't fit. I don't think his playoff averages alone are extraordinarily significant.

The poster said it has been statistically proven there is no such thing as clutch. I think these situations are indicative and clutchness could be inferred based on a common measure such as batting average.

All I really wanted to test was the probability of getting 7/8 hits when the mean was 2.61/8 (for a .326 hitter).

I'm a social studies guy taking a required course in which we do not do the math (thank God).

How out of the ordinary is it is my only question?

The probability of succeeding 7 out of 8 times, when your expected success rate is .326 is .0021, extremely low. The chances of Yaz hitting 7 out of 8 at bats at any point in time during the season is extremely low. The fact that this low-probability occurrence happened in a high pressure situation is where this debate comes from. It supports the clutch argument but doesn't prove it. I would love to know how many times Yaz went 7 out of 8 at any point during his career, it would be interesting to know how many times he achieved this specificly rare occurence, and we could broaden it to .800 over any period that included 6+ at bats or something.

Also note that this rare occurence (Yaz 7/8) is akin to flipping heads 35 out of 50 times.

 

[RayClay]

I understand that his results in the playoffs and world series were better than his "expected" results based on his career batting average. But this discussion is about where these "unexpected" results come from. How much luck is involved, how much "clutch ability" is involved, etc... We can compare numbers all we want but they will never prove one way or the other of WHY the numbers differ. The numbers tell us that in fact he hit better in the postseason, they don't tell us why, nor do they even suggest a reason for why. We can conclude that he WAS clutch, whether that was due to luck or some very rare ability to gain ability in higher-pressure situations, is not something we can conclude.

The clutch debate goes on because of the inability to prove it one way or the other, both sides have valid arguments and facts to back up the opinion of clutchness.

A batting average is a statistic.

If he got 7/8 hits in that weekend (.875 B.A,) and his average was .326 (2.61 hits per 8 at bats) he got more than triple the amount of hits to be expected.

That's a lot. And they needed both games.

Draw or don't whatever inferences you want.

 

[emoney_33]

A batting average is a statistic.

If he got 7/8 hits in that weekend (.875 B.A,) and his average was .326 (2.61 hits per 8 at bats) he got more than triple the amount of hits to be expected.

That's a lot. And they needed both games.

Draw or don't whatever inferences you want.

See above post. The likliehood of that happening is the same as flipping a coin 50 times and getting 35 heads. Does 35 heads out of 50 tries SEEM that crazy?

Yaz WAS clutch in those games, no doubt about it. But we are discussing the factors of being clutch. That is, did Yaz possess an ability to become amazing like that in the most necessary situations only, or was it a factor of luck or something else? We can't possibly know why this rare occurrence happened, we simply know that it DID.

 

[emoney_33]

See above post. The likliehood of that happening is the same as flipping a coin 50 times and getting 35 heads. Does 35 heads out of 50 tries SEEM that crazy?

Yaz WAS clutch in those games, no doubt about it. But we are discussing the factors of being clutch. That is, did Yaz possess an ability to become amazing like that in the most necessary situations only, or was it a factor of luck or something else? We can't possibly know why this rare occurrence happened, we simply know that it DID.

Also note how TINY of a sample size 8 at bats is. The difference of probability of getting at least x hits in 8 at bats changes tremendously by adding or subtracting ONE to x. 51% chance for at least 3 hits, 24% chance for at least 4, 8% chance for at least 5, 1.75% chance for at least 6, .2% chance for at least 7...

 

[RayClay]

The probability of succeeding 7 out of 8 times, when your expected success rate is .326 is .0021, extremely low. The chances of Yaz hitting 7 out of 8 at bats at any point in time during the season is extremely low. The fact that this low-probability occurrence happened in a high pressure situation is where this debate comes from. It supports the clutch argument but doesn't prove it. I would love to know how many times Yaz went 7 out of 8 at any point during his career, it would be interesting to know how many times he achieved this specificly rare occurence, and we could broaden it to .800 over any period that included 6+ at bats or something.

Also note that this rare occurence (Yaz 7/8) is akin to flipping heads 35 out of 50 times.

Good stuff. obviously you can't measure the reasons thing occur, only whether they occur more often than somebody else or an average.

I bet Yaz rarely had 7 for 8's as he was fairly consistent (though he changed his stance a lot). Having that stat would make for another interesting comparison.

I would maintain there is really no way to quantify clutchness in a kicker since actual clutch opportunities are so rare.

There's no doubt that Vinitraitor built up a lot of confidence over the years and the Patriots had confidence in him. He's not looking real confident these days (others noticed his look as well as I) for some reason.

All regular season kicks go out the window if a clutch situation arises. Some kickers go whole careers without many clutch opportunities.

 

[VJCPatriot]

would you care to make a wager on the Patriots going 16-0?

$50?

they go 16-0 I pay you $50, otherwise you pay me

So instead of addressing any of my arguments or logic in an intelligent manner you try to reduce this discussion to a tacky bet? Get some class, go back to actually READ the post, and come back with an intelleigent response. Thanks. For someone who was trying to make himself look smart, you came off with a really dumb and immature response.

 

[emoney_33]

Good stuff. obviously you can't measure the reasons thing occur, only whether they occur more often than somebody else or an average.

I bet Yaz rarely had 7 for 8's as he was fairly consistent (though he changed his stance a lot). Having that stat would make for another interesting comparison.

I would maintain there is really no way to quantify clutchness in a kicker since actual clutch opportunities are so rare.

There's no doubt that Vinitraitor built up a lot of confidence over the years and the Patriots had confidence in him. He's not looking real confident these days (others noticed his look as well as I) for some reason.

All regular season kicks go out the window if a clutch situation arises. Some kickers go whole careers without many clutch opportunities.

Yea I think we can all agree that AV was clutch for us. I don't care why he hit those field goals, he hit them and gave us great memories

 

[Fencer]

We can't possibly use this binomial distribution to prove or disprove clutchness right?

This test assumes the probability of the event is .285, but the argument of clutchness is that this probability rises in pressure situations. If we use this on the basis that a player cannot become better in pressure situations, then this only tells us that at least 24 successes out of 65 tries has a probability of .088, correct?


Using the binomial distribution in this discussion seems futile because it is operating under the pretense that "clutchness" does not exist.


To look at proof of clutchness or not, we must study all players' averages in "clutch" situations vs. all other situations. But this will only tell us that most players did not show much improvement over their norm in the clutch. The argument that CERTAIN players posessed this potential human quality can't really be proven or disproven based on their performance numbers. I'd guess it would be more a study of the human brain, affect of extra adrenaline, etc. etc...

Am I wrong here?

You're wrong on the numbers. The null hypothesis we're testing is whether or not the probabilities stay the same from nonclutch situations to clutch. If they do, then the binomial distribution is a good predictor of outcomes. THEREFORE, conversely, if the binomial distribution proves empirically to be a terrible predictor of outcomes, that's evidence against the null hypothesis.

You're in better shape on the substance, because of the great difference between proving correlation and proving causality.* It could be hard to separate out true clutchness from random factors.

*Stats can be used pretty easily to prove correlation or to disprove causality. However, it's much harder to use them to disprove correlation or to prove causality. (3 of the 4 immediately preceding claims should be obvious to anybody who's studied stats. What I mean by the 4th -- that stats can't easily be used to disprove correlation -- is that it's hard to prove a TOTAL lack of connectedness just because the data looks pretty random overall.)

 

[mb6592]

anybody who has studied Sabremetrics in baseball has learned a lot about how to analyze and predict future events. Theo Epstein has used these concepts to win 2 World Series and build an organization that can win year after year. The Cleveland Indians built the 2nd best team in baseball with a tiny payroll, and the whole team will be back next year - all built on sabremetrics. What lessons can we learn from baseball and use to analyze and predict football? Lots.

1) Variance happens and matters

Have you ever walked by a roulette table and noticed that the ball has landed on red 10 times in a row, and figured that the ball was "due" to land on black? Well, you're wrong. Sometimes weird streaks like that just happen, and there is no explanation needed other than to acknowledge that variance exists and manifests itself all the time in sports.

the most basic example? the idea of "clutch", which is generally highly overrated by most people. is David Ortiz a clutch hitter? most people in Boston would laugh at the question - of course he is, maybe the best clutch hitter of all time!! if this is you, then please please please read this article

http://sports.espn.go.com/espn/page2/story?page=betweenthenumbers/ortiz/060405

the conclusion, by Baseball Prospectus author Nate Silver, is that he is indeed clutch, but the ability is far far overstated.

in football, what does this mean? Well, read the Adam Vinatieri Mr. Clutch thread. my basic opinion is that AV makes the frozen ball kick vs the Titans or the blizzard kick vs the Raiders AT MOST 50% of the time. The odds of him hitting both? 25%. of course, he ended up making both. this doesn't make him clutch, it makes him a beneficiary of positive variance - instead of the likely 75% outcome (him missing 1) he got the 25% outcome. like a roulette wheel landing on red twice in a row.

now, I realize this is a foreign concept to most of you, so I'll give another example. This week AV missed a 29 yarder to win the game. Does this mean he has lost his magical clutchiness? Or that is he now a choker b/c he's on the Colts? No, neither - it just means that variance turned against him. even at his age, AV makes that kick 95% of the time. but THIS time he missed. too bad for him, funny for us. but please just recognize that AV benefited from variance in 2001 and 2003 - it wasn't b/c he was magically clutch.

given a large enough sample size, things tend to even out, which is what baseball analysis has taught us, and what AV is beginning to show.

2) The best team doesn't always win

Related to the above. the Patriots winning the SB 3 years in a row meant we were the beneficiaries of variance. even if you think the Patriots had a 70% chance to win each game, that means we win all 3 games 34% of the time. but we won all 3.

of course, there was no way we were 70% to beat the Rams. We were 14 point underdogs; we win that game MAYBE 15% of the time. but on that day we won. Ty Law had a ridiculous game, Mike Martz inexplicity stopped giving the ball to Marshall Faulk - these are things that don't always happen, but that day they did. sometimes the calls go against you (ie vs the Colts 2 weeks ago) sometimes key players get hurt at the worst times (the AFCCG last year), sometimes the ball bounces your way (almost the entire 2001 season), and sometimes the strangest, most unlikely timing means the game is not lost and you have a chance to win the game (Tuck Rule). teams don't plan these things, but they happen, and they are all reasons why on any given day the better team can lose. If the Pats lose to the Bills on Sunday b/c 100 things go wrong, everyone here will say that the Pats are the better team...but if you take a poll here, the majority of people will also say we were a better team than the Rams in 2001. learn to embrace variance guys, it happens and there is nothing you can do about it except build the best team possible and play the best you can.

3) our memories are awfully selective

again, related to the clutch discussion. most people here have no memory of the huge, clutch kicks AV has missed. the 2 kicks in the SB vs Carolina, the miss vs Denver in the playoffs, previous game winning misses. this is b/c they don't want to cloud their memory of the perfect clutch kicker.

another example: most people remember Tom Brady winning MVP of the SB in 2001. what most people don't remember is his actual performance in that game: 16 of 27 passes for 145 yards with a touchdown. Don't give me BS about "the gameplan calling for that". the gameplan certainly didn't call for Brady to complete only 59% of his passes, a below average mark. the gameplan didn't call for him to make 5.3 Yards/Attempt and only 92 yards with 2 minutes to go, and abysmal result. yes, he was awesome on the last drive, but before that he had completed 58% of his passes and had 4.8 yards/attempt - terrible. was the final drive great? Sure, but we were largely in that situation only b/c the offense and Brady had been so ineffective for the entire game leading up to that.

the lesson is that stats can give us a much more objective view of things. traditional stats tend to be bad, as they don't adjust for all kinds of things like strength of schedule, luck, and the effect one unit has on the other. Football Outsiders is the "Baseball Prospectus" of football, and every serious football fan should be reading this site. their team and unit stats DVOA and DPAR are far, far better than Points Scored, Yards Allowed, etc.

check it out: http://footballoutsiders.com/stats/teameff.php

Mark Twain said it best - There are lies, damn lies, then statistics. You don't need stats to prove sometimes you get the bear, sometimes the bear gets you, and sometimes some people have too much time on their hands.

 

[mb6592]

You can bet your bottom dollar any coach doesn't need a calculator to know the ball goes to Bird in the clutch (or maybe Garnett now). It's just another statistical method, there's nothing new under the sun and no method to replace a smart, experienced GM.

Exactly, but then what would these guys do with their calculators?

 

[mb6592]

But isn't this because their chance of something good happening is higher with the ball in the hands of Bird (or Garnett now), no matter the situation? Not just in the clutch, but in the clutch you can't make it up so with one chance you want your best player to have the ball.

And we need a statistician to tell us this?

 

[mb6592]

I love this post. You are correct, statistics never lie, but are rarely interpretted correctly and never perfectly. So nowww I'm going to bed .

The problem is not incorrect interpretation of data, statisticians are quick to through out numbers, but WILL NEVER tell you what the numbers definitively mean.

 

[mb6592]

Ok besides trying to be condesending and bringing a geek factor to Sunday football....WTF is the point to this thread? Really what is your point our definition of "clutch" isn't valid?

So we can't use "clutch" anymore right? Because statistics say so?

That's the funny point, statisticians will never say so. The data suggests, research indicates....

 

[mb6592]

Stats be damned. Anyone that thinks a kick or an at bat in the beginning of the game or in the regular season isthe same as in the end or win or go home situation has never played a sport higher than gym class or sat on the bench. It is simple as that. You are trying to use math class to quantify a situation it appears you know nothing about and you don't even realize it.

You can't see me, but I'm standing and applauding.

 

[makewayhomer]

The argument that CERTAIN players posessed this potential human quality can't really be proven or disproven based on their performance numbers. I'd guess it would be more a study of the human brain, affect of extra adrenaline, etc. etc...

Am I wrong here?

I think all the studies I linked to used individual data. if somebody was better in the clutch, the #'s would have showed. the amount of "clutchness" found was pretty minor

 

[makewayhomer]

The problem is not incorrect interpretation of data, statisticians are quick to through out numbers, but WILL NEVER tell you what the numbers definitively mean.

what are you talking about? I've linked to various articles and books in this thread which not only discuss #'s, but give their real world implications.