Glad you did that calculation! Now it's absolutely clear where the problem is. You're not doing the right calculation for the question that's being asked. OK.. look the question we're asking is this: what is the difference between the average win percentages for non-HoF QB's in close games vs. games that aren't close (or if one wishes, all games). We want to compare that to the difference between the average win percentages for HoF QB's in close games vs. all games. To answer that question, we need to look at each QB's win percentage (which is just 100 times the Pct column in that link), then average them in different conditions. You aren't averaging anything. You are just combining all wins and losses for all QB's and treating them as stats from a single QB. Neither Pauly nor I (nor anyone asking the stated question) is doing that or would do that. If you look at that Pct column and average all those numbers (and multiply by 100), you'll get 42% if you restrict it to QB's with 10+ starts and probably around 30% or maybe less if you take everyone. That's the correct calculation. Think about it. When Pauly reported not just a mean of 56.5% but a standard deviation of 5.24%, where do you think that standard deviation came from? Your calculation doesn't have a defined standard deviation. It's not the mean of any distribution. So yeah your logic is sound within a framework that doesn't matter for this debate, but the correct calculation for the question at hand is to average the win percentages for each QB, and that's where you get the logic I laid out.
i disagree..in hi pressure moments the mind can work differently and the steadiness of the hands can be affected by the moment..best example I have is Mike Irvin reiterating what I've always said when talking about pressure moments..he said with conviction.." I don't care who you are your going to feel the pressure so that's why you need to learn the art of catching the ball with your body, a slant pattern in a pressure situation I would catch with my body because my hands were not reliable in those situations"........ Pressure affects the mind and the hands imo..some folks get calm..some dont
No, that's not the correct calculation. The correct calculation is to divide all wins by a certain group of QBs by the total of all starts by that group of QBs. That's how you get the collective win% for that group. Your way gives meaningless and irrelevant data. As an example, if you have the following 4 QBs: A -- 200-92 (.685) B -- 194-60 (.764) C -- 0-10 (.000) D -- 35-35 (.500) The win% for that group is 68.5% (429-197), which is pretty outstanding. But using your method, the average win% would be 48.75%, which is subpar and falsely reflects the quality of that group. A second group of: E -- 20-180 (.100) F -- 20-80 (.200) G -- 9-1 (.900) H -- 9-1 (.900) has a collective win% of 16.7% (58-278), which is pretty pathetic, but using your method would have a win% of 52.5%, which appears to be better than Group 1. But clearly, nobody would claim that Group 2 is better than Group 1. Maybe that is how Pauly did it, but if so, then with all due respect that makes little to no sense.
That would be the case if you were just looking at the physical movements. But you cant tell me last play of the game....you make this catch and win..you drop it and lose...that nerves and confidence come hugely into play. The psychological aspects of running a slant on first down in the first Qtr....vs. running a slant with the game or even the season on the line are immensely different, IMO. Here's a great example of what I'm talking about. Hit a 9 iron shot to the green over a wide open fairway....nothing but grass. Now hit a 9 iron shot to the green....except there's a pond right in the middle. Same swing...same distance....but put water in between you and the green and its a much tougher shot.
By the time they reach the NFL and are playing a major role in big games, they have long ago learned to handle those moments. Not every time, of course, because they are human and they make errors and mistakes in clutch moments just like they do in non-clutch moments. Football has a lot of variables and small sample sizes (only 16 games per year, a limited number of touches per game, etc.) so its hard to come to clear conclusions as to what caused mistakes or great plays. But free throw shooting in basketball is a pretty pure no-variable exercise done under both pressure and relative non-pressure situations. And you'd be hard pressed to find any player who, over a large enough sample size, is a notably better (or worse) free throw shooter in pressure situations than in non-pressure situations. It really just doesn't exist. And that's doing something with no impact from other players, all eyes on the player and time to think about it (as opposed to just reacting). Sure, Jr. High kid might choke repeatedly in pressure situations at the free throw line, but you just don't see that (over meaningful sample sizes) from NBA players or even major college players.
Whether something is the right calculation or not depends on the question. If one asks what the average win percentage of a randomly chosen HoF QB is, then my calculation (and Pauly's, etc..) is correct. If one asks for the probability that a randomly chosen HoF QB will beat a randomly chosen non-HoF QB, then again my calculation is correct because you actually get distributions (with standard deviations), allowing us to see what the overlap is between the two groups. What you're doing gives no distribution whatsoever. What you're calculating is the answer to a different question. You're answering the question of what the percent of games that all HoF (or all non-HoF, as opposed to an average one) QB's have won, or are expected to win. I mean that's useful information but not for the question we're asking because ANY difference between two groups (even 0.1% difference) is statistically speaking significant because there's no defined standard deviation. There's a utility to having distributions so you can calculate the probability one group is different than the other. One can't do that with your approach. Of course as you point out one can get some results that are skewed doing what I'm doing because of small sample sizes. That's why one usually sets a minimum threshold for number starts to minimize that (why I used 10+ starts, which is probably a bit low). Anyway, the point is for the question we're asking the proper thing to do is look at distributions of win percentages, with some minimum number of starts.
I disagree though.... Tell me why Chuck Knoblauch grew up being able to throw the ball to first base, then all the sudden he couldnt. Joe Torre eventually had to move him to DH or outfield......its called the Yips. Heres one for football In 2007, Braylon Edwards was one of the most productive receivers in the NFL. He caught 80 passes for 1,289 yards with 16 touchdowns and made the Pro Bowl. The following season, however, Edwards production dropped nearly in half. Most notably, he led the NFL with 16 dropped passes including numerous wide open passes. He had a funny way of making tough catches in traffic, but could not reel in passes while wide open in stride. Heres one for basketball: John Starks was one of the best free throw shooters in the NBA during the 1995 season. However, Starks was struck with a case of the yips at the absolute worst time. In the midst of Reggie Miller's infamous eight points in 8.9 seconds in Game 1 of the Eastern Conferece Semifinals, Starks had the opportunity to deny the Pacers and Miller's last-second greatness. He was fouled with the chance to hit two free throws to ice the game. Starks clanked both and made Miller a legend.
I think you just inadvertently highlighted the problem with this stat/concept. What if the QB is clutch, but the WR isn't? Or the LT? or the C? Or someone on defense after the offense already put the team ahead? Or the coach? The simple fact of the matter is that people are talking about clutch as if players play better in given situations. Which is either not possible or the player is dogging it at other "non-clutch" times.
I understand what you are saying, but I guess for me your/Pauly's approach purports to answer a question that I find of little import or value. There is nothing wrong with not having a standard deviation due to using a collective and large sample. I think it is better and more meaningful.
None of those examples make or help make your point. Knoblach couldn't throw in any situation -- clutch or non-clutch. Edwards was dropping passes regularly, whether it was a clutch situation or a non-clutch one. Starks missed two free throws. It happens. Even to the very best free throw shooters. I've seen Dirk Nowitzky and Steph Curry do it and they are all-time elite free throw shooters. Starks was a 77% career free throw shooter, which is basically average. He misses roughly 1 out of 4 and in any distribution of 100 free throws for a guy like that they will likely miss two in a row a few times. Needless to say, two free throws is not a meaningful sample. Yes, there are important mental aspects to sports. Some guys lose it. But typically when they lose it they lose it. Not just in clutch situations.
OK, then let's just make sure that in the future the questions are precisely enough stated so that we can agree on the math that helps answer it
To me, arguing there is no such thing as clutch is arguing these games are being played by machines, and not humans. Just look at what the scandal did to Tiger, even before injuries. It ruined him. The effect was immediate. To a lesser extent, when someone wants to "force it" they make mistakes. More so than not under pressure. There are outliers and everyone else is in the middle.
I'm in the middle here. I think "clutch" definitely exists, but I think people make way too much of a deal about it in regards to judging players. The lists of "most clutch" are always some of the best players in the game. It's not like some player who isn't any good during the first 55 minutes is going to all of a sudden become a superstar because it's the end of a tight game. There's a reason why some of these lists are littered with hall of famers, and that's because they're hall of famers.
There's actually one fairly well studied property in psychology that would lend some credence to the idea of "clutch", and it suggests a precise mathematical formulation of it. Fitts's law is a mathematical relation that describes the "speed-accuracy" trade-off, and it works really well in a lot of situations (with some exceptions): https://en.wikipedia.org/wiki/Fitts's_law Basically, it says the average time taken to complete a task is a function of task difficulty where task difficulty in the original experiments was defined based on distance to, and the width of, a target object, but has since been extended to all kinds of different tasks. The mathematical relationship essentially says T = a + b*log(D) where T is time, D is difficulty (which has to be precisely defined for your preferred task), and "a" and "b" are parameters fit to each person. That logarithmic relationship works really well for a wide range of tasks (that is, the discovery was the logarithmic relationship). Point is this: "a" and "b" are person-dependent, meaning as the difficulty of a task increases (remember, difficulty here is defined objectively, not subjectively), people adjust to the increased level of difficulty at different rates ("a" tells you a baseline level of ability, and "b" tells you that rate). That's basically telling you that as you increase pressure, people respond at individualized rates to it even if their baseline level is the same. Yeah I know, it's not been tested for the precise situation we're talking about but Fitts's law holds really well in a huge range of tasks. This is probably the closest I think we can come right now to saying there's some quantifiable psychological evidence people in general behave differently under pressure situations.
There is no such thing as clutch, there is only playing at your best level or cracking under pressure. It is not possible for a person to play better than they are capable of playing at any other time. And since football is very much a team sport, W&Ls don't accurately tell anyone if a given player is clutch or not.
I agree...my point was that the mental part of the game can have a significant impact on performance. Some thrive in high pressure situations... Others let the high pressure affect their play.
By the way... I do think this is one of the better discussions we've had in awhile. Sports psychology is fascinating imo
"thriving" under pressure just means you under perform relative to your capabilities unless there's pressure and is therefore an undesirable trait.
That is an interesting point. It seemed for many years some nobody would end up staring in the World Series to help his team win.
Asafa Powell is a very important figure if you're trying to understand clutchness. He was the world record holder in the 100M dash, but only ever won a single semi-major gold medal in the event. He has never won a big race in his career, even while holding the title of "Fastest Man in the World". He was just never able to perform at a high level on the biggest stages. There's something there. I won't say it's clutch, but something prevented him from getting over that gold medal hump. Interestingly enough, he was a part of a number of gold medal relay teams. Relays can cut the pressure by a lot, and he never ran anchor from what I can remember.
But again, its not that Tiger just became bad in the clutch. He just became bad. It's not that they are machines, its that by the time these guys reach the pinnacle of their sport, they are used to playing under the lights and are used to pressure. They are playing, not pondering their legacy. On the fringes there may be some players who might tighten up and perform worse in certain circumstances, but there's no real evidence for anyone really being better in clutch situations. I come back again to free throw shooting because it is pure and without real variables. It is generally accepted that Michael Jordan, Kobe Bryant, Larry Bird and Dwyane Wade are "clutch." But none of them increased their FT shooting % in the playoffs, when the games are bigger and more meaningful, by any meaningful amount. Jordan's and Kobe's FT% dropped a bit in the playoffs. Bird's went up by 1.4% and Wade's went up by 0.9%. Similarly, Tracy McGrady, John Starks and Vince Carter have, rightly or wrongly, been labeled playoff chokers, but their FT% in the playoffs is within 1% of their career regular season FT% for each of them.
There was far more to Tiger's decline than the scandal. There were two or three back injuries, and all the surgeries, combined with doing two or three different swing changes.
But there aren't many examples of the guys who over a long period of time perform so different in pressure or non-pressure situations. In the few examples that do exist things even out as the sample size gets bigger, which is what one would expect whenever you have a player performing well above or below his usual norms.
Clutch is a media term used to pump up the favorite players. It sells games. "Big Shot Bob" Horry is a prime example.
Was it Rafael Furcal who blasted a home run to win a World Series for the Braves in the 90s? He was called "clutch" that game, but his career wasn't "clutch." Aaron Boone was "clutch"for one game, but his career didn't bear that out. Edit: Furcal was in Game 2 of the 2004 NLDS I always surprised when stat monkeys (lol, I love that term, no offense intended) buy into the "clutch" thing.
Of course, Horry's playoff shooting (FT, FG, and 3 pt) was virtually identical to his regular season shooting.
Hasn't Powell's career time period pretty much matched Usain Bolt's? It seems like Powell may have just been kind of unfortunate in that his prime largely overlapped with Bolt's.
Pretty sure its only Tannehill homers that don't believe in clutch or other intangibles. And that win/loss record is not a QB stat.
Just how important is "clutch", really? Well that's kind of what I mean. There's something to be said about average guys stepping up in big spots. It does happen, but more often than not those guys stepping up are the stars of the show. That's why I believe it exists, but it's not an elixir like some people make it out to be. It's an overrated stat, IMO.
Not really. Powell's first Olympics were in 2004, where he placed 5th. He had a couple World Championship tries before Bolt hit that stage, and before Bolt became dominant, he was losing to Bolt at speeds he had routinely bested in smaller races. Here's a telling quote from Powell on the 2007 World Games: I remember this race well because Derrick Atkins had surprised our whole country by pulling a silver medal, and proceeded to never do anything of note again.
Not a homer, but W/L isn't necessarily a QB stat. Baseball had a similar fixation on W/L but it has gradually waned as more and more people understand the impact that surrounding cast plays on wins and losses. You can only throw a ground out slider if your team can play defense. Same with pop-outs and other non-K outs.
Just how important is "clutch", really? Football is a team game. If win/loss records are a QB stat than why did hall of fame quarterbacks like Eli Manning, Philip Rivers, and Drew Brees lose so many games last year and in prior years? I'll hang up and listen.
Just how important is "clutch", really? And baseball pitchers deal with the same thing. Have you ever heard of Felix Hernandez? He's one of the best pitchers in the game, and has been for years. Now go look at his W/L record. Spoiler: it's not that great. Any smart baseball fan laughs at W/L records when judging a pitcher's overall performance. Too many other factors go into it. You know, because these are team games. If you want to judge an individual based on wins and losses, focus on golf and tennis.
If I'm understanding correctly what you are describing, I don't think it really applies or supports clutchness among professional athletes. In order to become professional athletes at the very highest level such that they even have an opportunity to be "clutch," these guys have already pretty well mastered the ability to adjust to changes in the level of difficulty. Also, the "clutch" situations we are talking about here aren't actually changes in the level of difficulty, nor are they things that have to be done faster or in less time. A free throw is the same task in the first quarter of the preseason game as it is in the final seconds of game 7 of the NBA finals. Catching a pass is the same task whether in the Super Bowl or on a practice field. But maybe I am again misunderstanding what you are talking about.
Where "speed-accuracy tradeoff" directly applies in football (say for a QB) would be for different levels of defensive pressure on a QB for completing the same pass. You have less time, or passing lanes are blocked, or you have to fit the ball in a tighter window, etc.. and you still have to complete the pass. Where the only change is the potential cost of failure ("pressure" situations), you'd probably define the increase in difficulty by saying the minimum threshold for success has been raised. So during the 1st quarter, it hurts but is not fatal if the QB tries to throw a TD but instead just gets a 1st down. Do that when you absolutely need a TD on the last play of the game and it's fatal. The question really isn't whether Fitts's law applies to this situation. It almost certainly applies. The question is what the distribution of parameter "b" (the rate at which you adjust to changing levels of difficulty) is for professional athletes compared to normal people, in the game they're professionals at. If the distribution of "b" has a narrow spread, then you're right that while Fitts's law applies, "clutch" is mostly a myth (there has to be SOME spread just due to random chance). If the distribution of "b" is as is normally observed and fairly wide for many tasks, then you'll see big differences among people. I don't know the answer to that question. One thing is useful though about Fitts's law: the mathematical relationship at least shows you how to quantify clutch or the lack of it.
There you have it, climate change is not a thing, its snowing outside. Notice, none of those guys have a losing win/loss record.
