it is a real thing as pros talk about it, before the pressure they are just playing, but when the pressure turns up they feel it
Yeah, but this idea that people play better than their norm when under pressure is ridiculous. Sure there can be more pressure, but, you get used to it. Like, in golf. When I first started playing, par putts were crazy hard, cause I wasn't used to them. Then par putts were no problem, but birdie putts made my knees knock. Now I'm getting more used to taking birdie putts, and those aren't such a big deal anymore. Or teeing off with people watching. Used to freak me out. Now, I don't care.
I have looked at the win% of QBs who have started a minimum of 100 NFL games and their record in 0-7 point games. Stat monkeys have long said that a record in 0-7 trends back to 50%. Since every game has a winner and loser this is correct for the league. From what I have found there are 86 QBs who have started 100 games that we have stats for their record in 0-7 point games. We don't have that information for pre 1960 games. Overall: 86 QBs. Win% in 0-7 point games: 52.3 This is a lot higher than expected considering this is a record of nearly 6000 0-7 point games. But of those 86 QBs 19 are in the HoF. I think any reasonable person will agree that Brett Favre, Peyton Manning and Tom Brady, who are not in the HoF yet will be first ballot HOFers. I will therefore treat them as if they were already in the HoF for analysis purposes. If we treat HoF QBs and non HoF QBs as separate groups. Hof QBs 22 QBs Win% in 0-7 point games: 57.1% Non HoF QBs 64 QBs Win% in 0-7 point games: 50.6% The non-HoF group of QBs performs basically as expected, very close to 50% and no evidence that 'clutch' is more than random chance. By making the cut off 100 starts we are making sure we are looking at long term starting quality QBs. The HoF group are performing far beyond what could reasonably ascribed to random variations. What this demonstrates is that some QBs have the ability to consistently win more close games than average QBs. When you consider that most QBs play an average seven or eight 0-7 point games per season winning 57% of them equates to winning an extra half a game per season. So to my mind it would be close to impossible to say whether a QB won any particular game because of their 'clutchiness', but it is more an overall effect that is in play whenever they are in a close game. What I think the debate is about is where that shift in win% above or below 50% comes from. What you describe is well known and I think it counts a long way to excluding execution of physical skills as the cause in the shift in win%. My belief is that it has more to do with decision making, the 'field general' side of being a QB - knowing which receiver to throw to, knowing when to eat the sack or to throw the ball away, making good down and distance judgements, knowing when to throw for receiver to get YAC and when to throw to help the receiver to get out of bounds. Some posters however believe that it is more in the execution of physical skills. The only way to get a proper answer is really extensive film analysis, PS When I started this I was firmly of the belief that 'clutch' did not exist. precisely for the reasons you say. It's only when I looked at the difference between HoF and non-HoF QBs that I realized there is something to be investigated further,
When people say clutch doesn't exist, they mean a player doesn't play at a higher level than normal in pressure situations. Of course a player can crack under pressure. But it seems just "not cracking under pressure" isn't what people are talking about when talking about clutch. It seems they are talking about a player playing better than they can....which doesn't make sense from a logic level.
Clutch is merely a euphemism for things such as maintaining focus. There has been research on this that indicates pressure can lead to a decrease of working memory. Sent from my SAMSUNG-SM-G920A using Tapatalk
Performance will always be relative to those a player is competing against, so someone playing better could merely be a player handling pressure better than those in the same situation. Sent from my SAMSUNG-SM-G920A using Tapatalk
Nate Keading destroys your whole argument. Clutch means performing under pressure. So if you're good, you're still good when the big lights are on. It doesn't necessarily mean a player gets even BETTER in high pressure situations. The bold part is exactly the point.
If a player can "crack" under pressure, then it stands to reason that a player can "elevate" under pressure, some people lose focus, others have their focus intensified. I can't speak for everybody, but if a player can maintain his own level under high pressure situations I consider that clutch.
Not really though to the bold. If the best a player can play is X, then they cannot exceed that by definition. If they are playing at level X in pressure situations and that's better then they normally play, then the level they normally play is <X. That means they aren't playing their best at other times essentially....which isn't a good thing. In this way, you can either play as good as you can play or worse. But I guess we're kind of saying the same thing.
Yeah it's important to understand what we're arguing so that debates aren't based on misunderstandings. As you point out people can crack under pressure. The question is whether a person cracks less under pressure than the average person in the same situation (like Stringer Bell pointed out). That's where the whole idea of looking at drop-offs comes from. Look at the average drop-off in whatever stat you care about (say 4th quarter performance from overall average, etc..) and compare the drop-off for your QB to the average drop-off. If statistically speaking that drop-off was unlikely to occur (quantifying what "unlikely" means by looking at a distribution), then you have statistical evidence for "clutch" in the sense that the person doesn't crack as much under pressure as others do. Of course, you'll have to look at many other such stats to see if there's a pattern or you're just cherry-picking one stat. There is of course the rare case of someone performing better under greater pressure. Flacco is the best example. After 2 disastrous playoff seasons, he just got better and better till he consistently averaged a higher rating in the playoffs than during the regular season, but that's rare of course: http://www.pro-football-reference.com/players/F/FlacJo00.htm
Adrenaline can push a person to go beyond what is normal, being in pressure situations can raise the adrenaline, high pressure can also "galvanize" a person and thereby raise his normal focus level. Same thing occurs in the Army, generally speaking, you can do more push ups with an angry drill sergeant behind you than you can on your own, which is based on fear. Bravery is "clutch" performance, people have in many instances performed well above their normal capabilities in critical situations.
Why are we surprised that HoF QBS have better win percentages in 0-7 point games than non HoF QBs? HoF QBS are better, hence, HoF, so...
I bet if you had to have an angry drill sergeant yell at you for many years while you did push-ups, the effect of the angry drill sergeant would become much less pronounced. You would get used to it, and would no longer fear it. You would eventually regress much closer to your mean. As the brain becomes accustomed to stressful situations, they become less and less stressful, thereby improving the level of performance.
"pressure" is a thing. That's why less than 1% of 1% of the population can play pro sports. Clutch, "the ability to play better at certain times", is not a thing.
To be honest I was. It's repeated as a mantra on stats sites that 0-7 point games regress to 50% and clutch does not exist. Before looking at the numbers I thought coaching would make a bigger impact. Also the correlation between passer rating and win% in 0-7 games is zero. So the Hof QBs aren't winning a higher% of 0-7 point games because they have a higher overall passer rating. Which makes sense because your passer rating is what helps you get into close games against evenly matched opponents. Passers with high passer ratings are playing their 0-7 point games against good teams and poor passers are playing their 0-7 point games against bad teams, so the result should be a coin toss. The specific nature of what HoF QBs are doing to move the win% needle in close games is the $64,000 question
Maybe I'm being simple minded, but I'm not surprised that better QBs do better in those situations than orise QBs do, regardless of opponent skill level. In other words, I would expect Tom Brady to make better decisions when playing the Seahawks than I would Mark Sanchez when playing the Browns. I don't see that as "clutch," but as better QBs doing what better QBs do. If the worse QBs made better decisions all the time, they wouldn't be worse QBs. HoF QBs aren't simply making good decisions in pressure situations, they always (well, not always cause they're not perfect) are making good decisions. So in pressure situations, they do what they always do. And the poor QBs do what they always do.
Eli Manning has had stretches in regular season where he was awesome. He's had times in playoffs where he's awesome. And he's had the opposite. People just remember the awesome. Same with Flacco
Flacco has had one bad game since his year in the playoffs in 2009. Since then he's 24 TD/4 INT and ridiculous ratings. Against better competition. Eli Manning in the playoffs has a better rating, usually where ratings go down in the playoffs because of tighter competition.
It is very important as that it is over in the coming down to the playoffs. As well as trying to go right down to the Nitty Gritty of It! Either win or go home!
I think any statistically significant evidence of "clutch" can be explained at least as easily by random variation or even just getting used to playing a defense over the course of a game as it can be by some special factor a player possesses.
That's precisely the reason why it's worth calculating the probability something is just due to random variation. Depending on the stat, we've seen stats that are less than 1% likely to be due to chance (correlation between Roethlisberger's <4 minute trailing stats and wins % or Luck's 0-7 win/loss record relative to all starting QB's in history, etc..) to 10% likely (Tannehill's <4 minute left trailing stats) up to 15-20% etc.. Of course, even for the 15-20% cases, you can argue it's more likely it wasn't due to chance because you have an alternative hypothesis for which the same data has a higher probability of occurring. Also, in most of these discussions, the effect of getting used to playing a defense or so is usually accounted for, not just because we're looking at population stats (so all QB's have similar chances to get used to the opponent), but also because it's not just one stat but many over which you might see a pattern for a particular QB. So sure you can't eliminate other hypotheses completely, but for many of the stats we're looking at, the most likely or one of the most likely explanations is arguably a trait of the QB.
There is actually a lot of scientific research on this subject: https://www.apa.org/pubs/journals/releases/xge-1304701.pdf https://hpl.uchicago.edu/sites/hpl....Pressure Multiple Routes to Skill Failure.pdf http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2708085/ Note - I haven't read most of that stuff, so not sure what conclusions they reach.
Looks like there are two dominant theories of why people can choke under pressure, according to those articles: "distraction theory" and "explicit monitoring theory". Distraction theory basically says you get distracted by other stuff, thus not performing as well as if you were not distracted, while explicit monitoring theory basically says you perform worse if you are forced to pay MORE attention to what you've already learned how to do well without being conscious of the step-by-step process you're using to do it. Apparently, there's a bunch of studies showing evidence for both theories in different situations. "Distraction theory" seems to hold when you need to learn or do something using working memory, so like a math problem where you have to keep individual steps temporarily in mind in order to get to the next step. "Explicit monitoring theory" seems to hold when you've learned something using procedural memory, like riding a bike or throwing a football, where you can do something without being able to explicitly identify each step of the process (that is, you'd have to think carefully about what you're doing to describe it even though it's easy to do after you learn it). So those articles are just saying there seem to be different ways you can choke under pressure, and that the two dominant theories, both with good amounts of evidence, seem to be either getting distracted from something that requires working memory, or having to pay more attention to something that you can do without explicitly identifying the step-by-step process.
HOF QBs winning a higher percentage of close games than non-HOF QBs is not surprising but also not evidence of clutch, IMO. Because they also win a higher percentage of non-close games. So even if everything in football came down to the QB (and it obviously doesn't) I would expect the team that wins 80% of their non-close games to also win a very high percentage of their close games.
Flacco's playoff career is 15 games, so less than 1 season. That is still a very small sample size. His career playoff rating is 88.6, which is hardly awe inspiring and not far from his regular season numbers. He's had some great games and he's had some stinkers. A partial list of some other QBs who looked really good in similar sized samples: Scott Mitchell Derek Anderson Nick Foles Josh McCown Brian Griese RGIII Etc. Eli's playoff rating of 89.3 is better than his career rating of 83.5, but his playoff "career" is disproportionately weighted by 2008 and 2012. It doesn't include all those bad years when he didn't make the playoffs or when his team lost in the first round of the playoffs. In those 2 seasons his regular season ratings were about 87, which is pretty close to that career playoff rating. And even for those 2 years, they are very small 4 game samples. The list of below-average QBs who have put together 4 game stretches similar to or better than Eli's 2008 and 2012 playoff runs is very large.
Yeah, so regarding resnor's and Fineas's argument about it not being surprising that HoF QB's have higher winning percentages in close games and that it's not evidence of clutch, I agree it's not surprising (the utility for me of Pauly's data was quantifying the difference, but I expected such a difference myself) but it is evidence of clutch, though I wouldn't argue it's as strong evidence as some other stats. The way I'd look at this is to assume each person's ability (or performance) differs across time because nobody behaves exactly the same way twice. HoF QB's will have one distribution of ability and non-HoF QB's another distribution. At any given moment in time, say for a specific game or specific play, you'll be randomly sampling from that distribution. Because HoF QB ability distribution has a higher mean, on average they'll win more and of course they'll win more non-close games. However, for there to be a close game, the randomly sampled abilities from the two QB's (assume just for argument's sake it's all on the QB) must have had approximately the same mean for most of the game. So there's no a priori reason to expect HoF QB's having better win % in close games just because they win more on average. The reason it wasn't a surprise for me is that you need similar QB abilities (let's say) to have a close game, but that doesn't mean HoF QB's in general don't respond better to pressure. And if they do, then in the crucial moments you'd expect better performance and thus higher win percentages. So it wasn't surprising for me, but it is (some, but not strong) evidence of clutch.
But that's the point -- you are cherry picking his numbers. That is like saying just look at RG3's rookie year and ignore the rest. If you ignore those early years then it becomes an even smaller sample size, just 10 games. In 15 games in March, the Heat's Josh Richardson shot 59% from 3. Needless to say, he is not a 59% 3pt shooter over any extended period of time, or anything close to it. It's a different sport, but the same premise applies. A 10 game sample is very small.
Well.. first of all it is still impressive that Flacco has a higher average playoff passer rating than regular season rating (88.6 to 84.7). Not many can say that. http://www.pro-football-reference.com/players/F/FlacJo00.htm Secondly, you're right it's a small sample size, but it will be similarly small with comparisons to other QB's, so while the variance is larger and the conclusions are thus less certain, the comparison can still be made for many QB's (excepting those QB's that have at most a few playoff appearances). Finally, a 15 game sample in the playoffs is like playing a single season. If the passer rating was for the most part very low in the first part of the season, but very high in the second part, that's probably something real and not due to random chance (if necessary I guess I could calculate this since we have the game logs but eyeballing it suggests there's a trend): http://www.pro-football-reference.com/players/F/FlacJo00/gamelog/post/
The fact that a game is close does not mean the QBs have equal ability or that they have played at an equal level. At all. One team may be close because of its running game and special teams and the other may be because of passing. Most times it will be a combination of many factors. Also, a close game may arise between a great QB and a bad one due to atypical poor play in the first 55 minutes by the great QB and atypically good play from the bad QB for the first 55 minutes. If Tom Brady's Patriots are only leading Cleo Lemon's Dolphins by 3 with 5 minutes remaining one would expect the Patriots to win for many reasons, among them: (1) they are already winning; (2) Brady has been playing poorly, with 2 INTs, but is reasonably likely to revert to his usual self over the last 5 minutes; and (3) Lemon is playing well but is unlikely to sustain that level through the end of the game, i.e., he is also likely to revert to his mean. That is without even accounting for the Pats' superior kicker (Gostkowski), better defense and better coach. Also, 7 pts doesn't mean much in terms of defining "close games." 46% of NFL games are decided by 7 points or less. In other words, that 7 pt margin is pretty close to the average NFL margin of victory. Also, looking at the win% in those games means little without accounting for who was winning going into he final minutes. Over the first 55 minutes, the better team will usually be winning. Say they are up by 7. As the better team with a 7 pt lead and only a few minutes left, that team is very likely to win the game. And the HOF QB of that team may have done nothing in those last 5 minutes other than hand the ball off and run out the clock to secure the 7 pt win.
That 4 pt difference in rating is very small. It is essentially equivalent to completing one extra 12 yard pass (not a TD or INT) in a 36 pass attempt game. That ain't much. And if it is so rare that so few QBs are better in the "clutch" by even that miniscule margin, it pretty well makes the point that essentially nobody is really "clutch."
Yeah I get all that. I was simplifying the argument by saying QB abilities are approximately the same in close games so that one could focus on why your logic was wrong. That is, there's no logic behind saying because HoF QB's win more non-close games, they should win more close games. They win more non-close games because the means of the distributions of team ability are different. That says nothing about what should occur if the means of team ability are similar, which is what happens for close games. And by replacing "QB" with "team", it annuls those other arguments about running game, defense, etc.. Point is, it's perfectly logical to expect 50% win percentage in close games even for teams that tend to win more because you only get close games if for THAT game the abilities of both teams were similar. Finally, the 7 point difference is probably a good one because it means you can tie the game with one score.
I agree it's small, but Flacco's gradual improvement in playoff passer rating to something that is much better than his regular season rating is probably real.
