Just how important is "clutch", really?

Then why wasn't the defense bad ALL the time?

 

Just off the cuff I would say some of that poor 3rd down percentage is caused by his avoidance of converting with his legs..he's a very strict pocket passer at this point..

 

and his inability to process what is going on.

 

Agreed, your counterargument actually supports cbrad's position.

 

I think it's pretty likely. One of the biggest, if not the biggest, factors that goes into a QB getting into the HOF is their W-L record and team success. There are a lot of HOF QBs whose own productivity/stats were relatively unimpressive but whose teams won enough that they got in as the putative leader of the team. So overall win percentage plays a huge factor in getting into the HOF for a QB. That typically also means winning a decent number or percentage of close games.

While I haven't taken a statistics class in many moons, IIRC this is called a dependent variable. Getting into the HOF is dependent, to a fairly large degree, on the QB's W-L record and overall team success. So seeing that HOF QB's tend to have higher winning percentages than non-HOF QBs is entirely unsurprising and is not necessarily related in any way to any reasonable definition of "clutch." Indeed, the difference of a 55% winning percentage for many (but clearly not all) HOF QBs and approximately 50% for non-HOF QB's seems pretty miniscule.

 

I think it was because they were predictable in that they would go to prevent/shell defenses late in games and would lose the aggressiveness that they showed in normal situations.

 

yeah he doesnt forsee where the pressure is coming from and have instinct to manuever the pocket with swiftness and purpose..

it should be his main offseason task.

 

And again, those other QBs have means Thill does not to counter the bad oline like a running game or being allowed to audible, for example.

That's the problem with these stats analysis, they operate under the guise that only what has been quantified is all that matters. The issue is, there are things not yet quantified that do matter, like being able to audible.

QBs have tools besides their own ability to deal with a poor oline, and most of those were taken from Thill. I know many want to blame the fact those tools were taken on Thill, which is a separate argument. This all boils down to is, (what I keep harping on and no one can seem to answer)....list the QBs that have succeeded countering a poor oline while not being able to audible and running the fewest amount of times in the league. If that question can't even be dealt with, the whole statistical argument breaks down, because there are huge pieces of information being left out.

None of these stats show anything to be true 100% of the time, so it makes sense to start thinking about what is happening in those times when its not 100%.

 

other qb's have skillset means as well..I would like to know if ryan in year 5 has those things..

 

I know they do. Most don't have them however. If they were faced with thill's issues their numbers would decline, not as far down as thill's but they wouldn't look all world either. I know you don't agree, but you're wrong.

 

It's hard to predict where to go when you know there will definitely be pressure coming from both ends of the line, and a good chance the interior was going to get pushed back.

 

Really? He's saying we need stats for 3rd down length in the 4th, and I'm saying there that we already know we were in longer 3rd downs than everyone else, went would that be different in the fourth? He's assuming that oline is the same in the fourth, and defense is the same in the fourth, why can't we assume or third down distances stayed the same in the fourth?

 

Yeah there are two separate questions here: 1) is that 55% really a big difference from 50%, and 2) was the greater winning percentage for HoF QB's due to the QB or to the team or coach etc..?

I don't have all the raw data, but if we go by Pauly's stat of 56.5% mean and Std Dev of 5.24%, the probability that a randomly chosen HoF QB will have a better W/L record in close games than a non-HoF QB is just over 80% (assuming the Std Dev for the non-HoF QB distribution is also 5.24%). That might seem high, but keep in mind the standard deviation is fairly small. So those two distributions have a decent degree of separation (btw.. 80% isn't as high as it might seem at first sight because 50% is chance, so keep that in mind).

To the second question, I have no idea but you definitely bring up a good point. Even if the two distributions are fairly well separated, it doesn't mean the QB was the reason for that.

 

Yes really. He/We are assuming the third down distance stayed the same in the 4th. See, this really shows maybe you don't see the whole argument. He agrees with you. Assume the third down distance is the same in the 4th, as it is during the first 3.

Since that is a constant, it is NOT a factor in this hypothesis. If you had stats showing 3rd down distance was indeed even longer in the 4th, that would help explain Tannehill's massive drop in rating.

 

That is the art Res, multitasking with your eyes, The qb position is about great peripheral vision and seeing things as a whole thn minimizing the picture from there in fractions of seconds..You've got to be able to keep an eye on your lineman and sense whos getting beat to be able to make your move in time.

 

One thing that would help understand these numbers is how their close game win% compares to their overall win%. For example, IMO a close game win% of 60% doesn't signal clutch if that same guy has an overall win% of 70%. I don't know for sure, but I'd guess that there are very, very few HOF QB's with sub-.500 records overall. Joe Namath might, but I suspect he's pretty close to .500. Fouts and Moon are very close to .500, but just above. Not sure who else, if anyone, is in that category. I'd guess that the average win% for HOF QBs is well over 60%, which may actually suggest that their average clutch win% of 56% or so arguably reflects a lack of clutchness. I don't really believe in clutchness (overall and in a large enough sample size) so I don't think its that.

 

Good question. Not hard to find the data for because everyone except Sammy Baugh is here (and he was pre-NFL era):
https://en.wikipedia.org/wiki/List_of_most_wins_by_a_National_Football_League_starting_quarterback

HoF QB's win an average of 61.4% of their games, with a Std Dev of 8. So if Pauly's stats are correct, win percentage decreases by 5 percentage points (8% drop from normal) in close games. Interesting that the Std Dev drops in close games though.. basically suggests HoF QB's play more similarly in close games than in games that aren't close.

Regarding the "clutchness" question, you'd have to compare that 8% drop to what all QB's do, and I don't have the stats for that. If that 8% is much less than non-HoF QB's, then it's "clutch", if it's similar then "no clutch", and wouldn't it be amazing if the drop was more haha!

 

What isn't measured in these stats, is "must win" games which include playoffs, and division games, because that is when the "clutch" QB is most heavily graded.

No offense to Pauly who does a great job with these things, but some very particular stats are extremely hard to come by.

 

We know that for all QBs the average win% is 50%. If HOF QBs have an overall win% of 61.4% of their games, then we know that non-HOF QBs must have an overall win% a little below 50% (there just aren't that many HOF QBs).

We similarly know that the average win% in close games is 50% (one wins and one loses, ignoring the miniscule number of ties). So if HOF QB's are winning 56.5% of their close games, non-HOF QBs must be winning just under 50% of their close games. In other words, non-HOF QBs would have no noticeable decrease in win% in close games, whereas HOF QBs have a larger dropoff.

 

I don't think division games are notably or necessarily more "must win" or clutch. Over the last 16 years or so, unfortunately, the Patriots' division games have been virtual gimmes.

Playoff game stats are readily available. You generally don't see even HOF QBs playing better in the playoffs. For example, Tom Brady's passer rating drops from 96.4 in the regular season to 88 in the playoffs. Peyton Manning's dropped from 96.5 to 87.4 in the playoffs. Dan Marino's dropped from 86.4 to 77.1 in the playoffs. Etc. Those drops may be due to quality of competition, weather, etc., but it certainly doesn't suggest clutchness. Joe Montana's playoff rating is a little higher than his regular season rating, but not enough to be statistically significant (I would guess).

 

Nice! Gotta give kudos when people engage in deductive reasoning like that!

Only one problem with your argument: it's almost certainly wrong haha.

Why? Because technically speaking what you said is exactly true if all QB's play the same number of games. If they play approximately the same number of games your argument will be a good approximation. But if they don't play similar numbers of games, your argument no longer holds.

Consider: you have two QB's A and B. Each started 10 games and combined their record is 18-2. You can distribute those 2 losses in any way you want (A gets both losses, or each gets just 1 loss, etc..) and the average win percentage will always be 90%. The reason: the denominator of 10 cancels in the math.

But if they don't play the same number of games, that's no longer true. For example, if A played only 2 games, lost both, while B played 18 games and won all 18, the average win percentage is suddenly 50% (extreme case of course). Or if A's record was 2-0 and B's was 16-2, the average win percentage is 94%. Point is, when the number of games played differs among QB's what you said is no longer true. Not only that, your argument becomes less and less of a good approximation the more non-HoF QB's you have that started relatively few games compared to HoF QB's.

And HoF QB's tend to have longevity. So I'd actually be surprised if "ability" isn't reflected in the stats. My guess is either the percent drop from normal for non-HoF QB's is around the same or worse than for HoF QB's.

 

Five Thirty Eight tried to tackle this clutchness.

http://fivethirtyeight.com/features...eason-quarterback-of-all-time-is-eli-manning/

(as of Jan 2015)




http://fivethirtyeight.com/datalab/tom-bradys-statistical-place-in-the-pantheon-of-nfl-qbs/

Measuring any/a during the playoffs these are career leaders (of course, clutch could mean how it compares to their regular season performance, but clutch sometimes means maintaining your good play, not exactly elevating it)

 

The good and the bad about that:

The good: they actually used some metric (Elo ratings of the teams at the time the match was played) to calculate how likely or unlikely it would have been for that team to have the postseason record they ended up having. Importantly, Elo is a predictive stat (you use a stat based on games played before the match to predict the outcome of the match) and not post-hoc.

The bad: 1) it's a team stat in the purest sense so it's not trying to measure QB ability, and 2) Elo is really way too widely used for it's (lack of) utility. Elo is just meant to be a computationally simple approximation of "true" estimated ability that really only works in a small subset of cases. It takes a long time for methods like this to be replaced but someday Rasch analysis will replace it. Just a question of inertia right now.

 

Why are we assuming that everything stays the same, except for QB play? If the QB experiences a large aberration from his norm, then it would stand to reason that other factors are coming into play, other than his ability.

 

This brings up the issue of what is the definition of "clutch". FiveThirtyEight goes with playoff performance, while your definition is on a per game/opportunity basis. One can argue either definition considering all playoff games are win or go home while 'crunch time' can exist even in the preseason. Metrics of playoff clutch-ability, however, are a bit sketchy IMO. Playoff appearances are a team accomplishment encompassing supporting cast offensively, defense, coaching, and random luck. I think I prefer the game by game metrics because of larger sample size, accessibility, and the lower influence of team factors.

 

Your complaint isn't really valid when you consider the sample sizes for both the HOF games and non-HOF games. We aren't talking about a comparison of two players who played 20 combined games. We are talking about a HOF pool that played thousands of games collectively and a non-HOF pool that played tens of thousands of games collectively. So we are comparing two groups of QBs, not individual QBs, and there is no sample size issue with either group.

Yes, HOF QBs typically have longevity and ability is definitely factored into their HOF status. Winning record played a role but most HOF QBs were really, really good. Non-HOF QBs, as a group, are much, much less good, but as a group their clutch win% drop is negligible (if it exists at all) and the drop for HOF QBs is not negligible according to the OP. In other words, the fact that the non-HOF group is littered with scrubs who may have only started a few games actually makes my point stronger.

 

Again, this ignores the fact that win% and HOF status are dependent variables for both QBs and coaches. QBs with bad win% simply don't make the HOF regardless of stats because they have bad win% and are not perceived as winners. Similarly, coaches with bad win% will never make the HYOF because winning is the #1 thing, by a country mile, on which they are judged for HOF purposes. Thus, your argument on this point becomes circular and self-fulfilling.

 

Actually going conservative is not going to inherently affect passer rating negatively. A bunch of completed, short passes is going to create a good rating. Accuracy is counted twice, in the ypa factor, and the completion percentage. Also INTs.

 

Oh it's absolutely valid. We're not combining the number of games. You're averaging records for whatever number of games the QBs played. So suppose HoF QB's play on average 150+ games in their career, while with non-HoF QBs you get many that play only several dozen while many others play 150+ or so. You'll absolutely get the effect I'm talking about.

Here.. I'll prove to you something that is counter-intuitive (well not to me). You'd argue the average winning percentage across all QB's is 50%. That can't be anything close to the truth based on my argument (has to be far less). So take all QB's on this list:
http://www.footballdb.com/stats/qb-records.html?type=&alltime=1

that started at least 10 games. Average their W/L records. Result: 42%.

Point is, number of games played will change the average W/L record across the league. It is NOT 50% nor anywhere close. And since in general you play more games the more you help win, the average for non-HoF QBs will be far lower than 50%.

 

Conservative, for many teams, involves running the ball to kill clock, not throwing a bunch of short passes, though.

 

No, because we aren't comparing individual players, we are comparing groups of players. The argument isn't whether Joe Montana was more clutch (or has a bigger/smaller dropoff in win% in close games as compared to non-close games) than Cleo Lemon, it is whether the class of QBs like Joe Montana (here, HOF QBs) has a bigger/smaller dropoff than the class of guys like Cleo Lemon (here, non-HOF QBs). Of course, the non-HOF class contains a lot of QBs much better than Cleo Lemon, but let's for a moment assume it didn't. Let's assume that class is all journeymen QBs each with 20 or fewer starts. But collectively, that group has 1000 starts. If that group wins 45% of their non-close games and 45% of their close games, then from the logic of the OP they would be neutral -- neither clutch nor non-clutch. But if the elite HOF group has an overall win% of 62% in non-close games, but only 55% in close games, then that group would actually be performing worse (if winning is the standard) in close games than in non-close games. I don't see how you can call someone clutch if they perform worse in close games than in blowouts. Of course, as discussed previously, win% isn't the only, or even the best, measure of performance or clutchness.

 

See the rest of the previous post I just added (sorry for the late edit). And yes we're asking whether an average (or randomly picked) HoF QB was more clutch. What other question would even make sense?

 

That's my point. You run the ball guess what, no affect on rating. Short pass on 3rd and 4 or 3rd and 5 etc. of course if you miss it, you take a hit on rating but that's not because of the conservative nature, it's because you missed the pass (rating doesn't know if it's a 3 yard dump pass or 50 yard bomb that was incomplete). Going conservative doesn't mean you take a hit on rating. Handing off the ball, has no effect on rating (obviously).

 

Of course the average winning % is 50% (excluding ties, which really shouldn't change anything). Two QBs start every game. One of those two wins and one of those two loses. 50%.

Not sure why you are taking only those that started 10 games. There's no reason to exclude the others. If the 10+ start guys have an average win% of 42% (I'll take your word for that -- I haven't done the calculation), then the less-than-10-starts guys must be well over 50%. I'd be very surprised if that was really the case, as it would suggest that those scrubs are winning at a much higher rate than the 10+ start guys.

 

Why don't you just look at the link to see your intuition is wrong:
http://www.footballdb.com/stats/qb-records.html?type=&alltime=1

Look at all those QB's that started less than 10 games.. or those that started less than 50 games. Just scroll through the list. For those that started 1 game only, the most common win percentage is 0%!! If we include everyone I bet the average win percentage is less than 30%.

Point is your intuition is wrong about this. If a randomly chosen HoF QB wins X% of close games on average, and many of those are against non-HoF QB's, then those close games for non-HoF QB's will have an outsized effect on their overall close games W/L percentage because you have to divide by the (smaller) number of games they play. The intuition should actually be that the drop-off from normal to close games win percentage is greater for non-HoF QB's (mathematically, that's the more likely outcome) because they don't play as many games.

 

It's not intuition, it's logic. For every single time a QB wins a game, another one loses it. And vice-versa. Who are you contending are losing all the games that other QBs have won, if not other QBs?

Yes, one would theoretically expect a bigger drop off from normal to close games for non-HOF QBs. At least that was the OP's premise. But the numbers that have been presented do not show that. They seem to suggest the opposite. I have not double-checked those numbers/calculations.

I did look at your link. And I did count up the wins and losses for current QBs, as it was a much smaller list than all QBs going back to 1991. Current QBs with 10+ starts won 2202 games and lost 1637 games, a 57.3% win%. Since current QBs excludes Peyton Manning, who dished out a lot of losses over his career, that number may be a bit conservative. In other words, if Manning was included it would have been much higher. QB's with less than 10 starts have a win% well below 50%, as expected. It's a pretty short list, as not too many of those guys are still in the league.

I am not going to do all the math for all the QBs, but I do not believe those with 10+ starts have a win% of 42%. Logic says otherwise and just eyeballing the list similarly tells me that isn't true.

 

Right, but a QB is also just as likely to throw the ball away or throw it away from a receiver, in order to avoid a turnover, late in games. Going conservative isn't simply throwing short passes as a QB.