Ryan Tannehill

"cbrad wrote:

Well.. here are Brady's and Wilson's year-by-year data points plotted against Tannehill's 2019:

You can see that the trend line is pretty good in predicting the distribution of the stats for both QB's. In other words, both those QB's (and I suspect others too) tend to perform around that trend line over time. So based on the single data point in 2019 Tannehill is doing slightly better than expected even for QB's like Brady and Wilson.

In any case, no one should mistake Tannehill for QB's like Brady and Wilson that have proven they are some of the best year after year, at least not until he proves he can do the same. But at least for 2019 there's no statistical basis for the claim that the correlation between his passing attempts and passer rating is somehow worse than would be expected for similarly performing QB's.

Again appreciate the work as always. Let me see if I can break down the 2019 season in a more understandable manner.

Consider the following numbers of pass attempts and passer ratings in games for a single quarterback:

14 -- 109.5
18 -- 155.8
19 -- 133.9
20 -- 130.8
22 -- 131.2
27 -- 140.4
27 -- 133.6
29 -- 120.1
33 -- 109.8
36 -- 92.2
39 -- 82.3

The correlation between pass attempts and passer rating above is -0.68, which is fairly strong.

Based on the appearance of the data, the question becomes, why does this quarterback play so much more poorly in high-volume passing games? His average passer rating in low-volume games is far higher than it is in high-volume ones.

Now let's insert two more games of pass attempts and passer ratings for the same quarterback:

15 -- 61.0
16 -- 78.1

What we have above of course are two low-volume games with low passer ratings. The insertion of those games, however, diminishes the correlation between pass attempts and passer rating from -0.68, which is fairly strong, to -0.089, which is almost nothing.

The correlation was diminished not by his playing well in high-volume games, but by his playing poorly in low-volume ones.

However, the quarterback's average passer rating in low-volume games is still far higher than his average passer rating in high-volume games.

Why should we not continue to question why this quarterback plays so much more poorly in high-volume passing games, especially when he has exhibited the same pattern of performance over his much longer career as a whole?

The above are the actual data for Tannehill's 2019 season by the way.

"The Guy wrote:

Again appreciate the work as always. Let me see if I can break down the 2019 season in a more understandable manner.

Consider the following numbers of pass attempts and passer ratings in games for a single quarterback:

14 -- 109.5
18 -- 155.8
19 -- 133.9
20 -- 130.8
22 -- 131.2
27 -- 140.4
27 -- 133.6
29 -- 120.1
33 -- 109.8
36 -- 92.2
39 -- 82.3

The correlation between pass attempts and passer rating above is -0.68, which is fairly strong.

Based on the appearance of the data, the question becomes, why does this quarterback play so much more poorly in high-volume passing games? His average passer rating in low-volume games is far higher than it is in high-volume ones.

Now let's insert two more games of pass attempts and passer ratings for the same quarterback:

15 -- 61.0
16 -- 78.1

What we have above of course are two low-volume games with low passer ratings. The insertion of those games, however, diminishes the correlation between pass attempts and passer rating from -0.68, which is fairly strong, to -0.089, which is almost nothing.

The correlation was diminished not by his playing well in high-volume games, but by his playing poorly in low-volume ones.

However, the quarterback's average passer rating in low-volume games is still far higher than his average passer rating in high-volume games.

Why should we not continue to question why this quarterback plays so much more poorly in high-volume passing games, especially when he has exhibited the same pattern of performance over his much longer career as a whole?

The above are the actual data for Tannehill's 2019 season by the way.

Hasn't this already been answered NUMEROUS times? Dude... seriously.... time to come up with something else.

You cannot remove games to increase the correlation..... that is cherry picking.....

How about we throw out the Carolina game. It was his only 2 INT game of the year. Clearly an outlier.

Let me suggest another theory. Tannehill had three games with more than 30 passing attempts.

33 -- 109.8
36 -- 92.2
39 -- 82.3

In two of the games, the Titans fell behind by double digits (17 points and 14 points). In the other, they never trailed by more than 6 and led the game for large parts of it. Without looking, guess which games are which.

"The Guy wrote:

Again appreciate the work as always. Let me see if I can break down the 2019 season in a more understandable manner.

Consider the following numbers of pass attempts and passer ratings in games for a single quarterback:

14 -- 109.5
18 -- 155.8
19 -- 133.9
20 -- 130.8
22 -- 131.2
27 -- 140.4
27 -- 133.6
29 -- 120.1
33 -- 109.8
36 -- 92.2
39 -- 82.3

The correlation between pass attempts and passer rating above is -0.68, which is fairly strong.

Based on the appearance of the data, the question becomes, why does this quarterback play so much more poorly in high-volume passing games? His average passer rating in low-volume games is far higher than it is in high-volume ones.

Now let's insert two more games of pass attempts and passer ratings for the same quarterback:

15 -- 61.0
16 -- 78.1

What we have above of course are two low-volume games with low passer ratings. The insertion of those games, however, diminishes the correlation between pass attempts and passer rating from -0.68, which is fairly strong, to -0.089, which is almost nothing.

The correlation was diminished not by his playing well in high-volume games, but by his playing poorly in low-volume ones.

However, the quarterback's average passer rating in low-volume games is still far higher than his average passer rating in high-volume games.

Why should we not continue to question why this quarterback plays so much more poorly in high-volume passing games, especially when he has exhibited the same pattern of performance over his much longer career as a whole?

The above are the actual data for Tannehill's 2019 season by the way.

Believe it or not a t-test is significant for the low- and high-volume passing games above, even though the high-volume passing games are an N of just 2. The reason of course is because the mean of the low-volume games (118.6) is so discrepant from that of the high-volume games (87.3), and there is comparatively very little variation in passer rating in the high-volume games.

"The Guy wrote:

Again appreciate the work as always. Let me see if I can break down the 2019 season in a more understandable manner.

Consider the following numbers of pass attempts and passer ratings in games for a single quarterback:

14 -- 109.5
18 -- 155.8
19 -- 133.9
20 -- 130.8
22 -- 131.2
27 -- 140.4
27 -- 133.6
29 -- 120.1
33 -- 109.8
36 -- 92.2
39 -- 82.3

The correlation between pass attempts and passer rating above is -0.68, which is fairly strong.

Based on the appearance of the data, the question becomes, why does this quarterback play so much more poorly in high-volume passing games? His average passer rating in low-volume games is far higher than it is in high-volume ones.

Now let's insert two more games of pass attempts and passer ratings for the same quarterback:

15 -- 61.0
16 -- 78.1

What we have above of course are two low-volume games with low passer ratings. The insertion of those games, however, diminishes the correlation between pass attempts and passer rating from -0.68, which is fairly strong, to -0.089, which is almost nothing.

The correlation was diminished not by his playing well in high-volume games, but by his playing poorly in low-volume ones.

However, the quarterback's average passer rating in low-volume games is still far higher than his average passer rating in high-volume games.

Why should we not continue to question why this quarterback plays so much more poorly in high-volume passing games, especially when he has exhibited the same pattern of performance over his much longer career as a whole?

The above are the actual data for Tannehill's 2019 season by the way.

Dude.. even for cherry picking that's showing lack of skill. You leave in a game with 14 passing attempts but take out two with 15 and 16 passing attempts to make a point about low volume games? C'mon.. cherry picking can be done better than that!

Anyway, you can't remove data like that if you care about proper statistical methodology, and I've already shown the results of using proper methodology. So my role here is over I think lol.

This dude...…...LOL

Whatever the Titans are doing defensively in the red zone, the Chiefs have to solve. With respect to Henry, whatever the Chiefs have in the form of thumpers at LB & run-stuffers along the DL, they have to be out there on early run downs. It's tough because those guys are often liabilities vs pass. We may see them play a lot of press/jam on the outside or all 11 in the box. Perhaps some 5-man defensive fronts aggressively firing off the snap in one direction or the other. Quality hits on Henry in the backfield is how you mitigate the power & vision of a RB like him. Well timed run dogs, Henry's a long-strider, so unanticipated adjustments in the backfield are big too.

"PhinFan1968 wrote:

And the hits keep comin'! Tannehill has been voted the Professional Football Writers of America's "Comeback Player of the Year" and "Most Improved Player."

https://www.paulkuharsky.com/news/ryan-tannehill-doubles-up-on-pfwa-awards

"He became just the third quarterback in NFL history to post a completion percentage over 70 and a yards per attempt over 9, joining Joe Montana and Sammy Baugh.

A decisive thrower, Tannehill has thrived in play-action situations. He led the NFL with a 117.5 passer rating and 9.6 yards per attempt.

With six games with a passer rating of 130 or higher, he tied Russel Wilson (2019), Tony Romo (2014) and Aaron Rodgers (2011) for the most in a season since 1960. Four of those games came consecutively, against the Chiefs, the Jaguars, at the Colts and at the Raiders."

And in the presser where Tannehill had just found out about the awards, as usual, he was humble and respectful in his reply:

"I think the awards are just a reflection of the guys that we have on the team."

"PhinFan1968 wrote:

And in the presser where Tannehill had just found out about the awards, as usual, he was humble and respectful in his reply:

"I think the awards are just a reflection of the guys that we have on the team."

Craziest thing about this run Ten has had since Tannehill took over is you would think they would dominate time of possession. Its been exact opposite. They have only had more time of possession in 2 of their 9 wins and are around 27th in league over that time.

"cbrad wrote:

Dude.. even for cherry picking that's showing lack of skill. You leave in a game with 14 passing attempts but take out two with 15 and 16 passing attempts to make a point about low volume games? C'mon.. cherry picking can be done better than that!

Anyway, you can't remove data like that if you care about proper statistical methodology, and I've already shown the results of using proper methodology. So my role here is over I think lol.

I don't think you're following, because the point is about high-volume games, not low-volume ones, but I suspect you may be busy with other things. Again the point is that in 2019 there have been NO high-volume games that were played well, while there WERE low-volume games that were played poorly.

Obviously that configuration of data is going to diminish the correlation between pass attempts and passer rating considerably, while preventing a correlation from being a valid indicator of the fact that NO high-volume games were played well. In other words, a correlation is a distorted representation of the data.

The question isn't "can Tannehill play both well and poorly in low-volume games." The answer to that question is a clear yes. The question is, "why does his performance diminish so much more than that of QBs with similarly high overall passer ratings in high-volume games."

Correlation is the wrong measure in this case. But you've put in a lot of work here (as have I) and it appears we've reached an impasse, so thanks.

"The Guy wrote:

I don't think you're following, because the point is about high-volume games, not low-volume ones, but I suspect you may be busy with other things. Again the point is that in 2019 there have been NO high-volume games that were played well, while there WERE low-volume games that were played poorly.

Obviously that configuration of data is going to diminish the correlation between pass attempts and passer rating considerably, while preventing a correlation from being a valid indicator of the fact that NO high-volume games were played well.

The question isn't "can Tannehill play both well and poorly in low-volume games." The answer to that question is a clear yes. The question is, "why does his performance diminish so much more than that of QBs with similarly high overall passer ratings in high-volume games."

In other words, correlation is the wrong measure in this case. But you've put in a lot of work here (as have I) and it appears we've reached an impasse, so thanks.

Maybe because of the scheme? Maybe beacause WRs cant create seperation? Theres alot of things that factor into those things.

"Cashvillesent wrote:

Maybe because of the scheme? Maybe beacause WRs cant create seperation? Theres alot of things that factor into those things.

But now consider that we're talking about a pattern that's existed over Tannehill's seven-year career as a whole, across many configurations of surrounding variables.

"The Guy wrote:

But now consider that we're talking about a pattern that's existed over Tannehill's seven-year career as a whole, across many configurations of surrounding variables.

Stafford has played in an offense of the DET that was very high volume, so has Andrew Luck with the Indianapolis Colts, how would you rate those QBs?

"Cashvillesent wrote:

Stafford has played in an offense of the DET that was very high volume, so has Andrew Luck with the Indianapolis Colts, how would you rate those QBs?

Interesting question. Stafford's career average passer rating in low-volume games is 97.2. In high-volume games it's 85.7.

For Luck those figures are 96.3 and 88.8.

Again for Tannehill those figures are 97.9 and 80.5. Once again his figure in high-volume games is a good bit lower (though less in these cases) than even quarterbacks people might consider average (as opposed to the earlier comparisons with Brees, Mahomes, etc.).

"The Guy wrote:

That would be plausible hypothesis if he didn't demonstrate the same pattern of performance this year, where his passer rating plummeted, on average, from 118+ to 87.3 in high-volume games.

And again that degree of performance decrement between low- and high-volume games isn't what we see with the league's top QBs.

Let’s look at these 2 datapoints
First you havethe 20-30 loss to Carolina. Tannehill had an 82.3 passer rating. Good to win about 40% of games if I’m doing the math right. Carolina got out to a 17 point lead in the first half and forced the Titans into pass first mode. Few QBs have good games once they get down 0-17. Game recaps said that the Titans gameplan did not establish Derek Henry and the run in the first half.

Second was the first game against Houston. Tannehill had a 92.2 rating, good for 50% chance of winning the game. However ne play in particular stands out and it was discussed at length at the time. Tannehilk threw an 18 yard pass to the 1 yard line which was caught by the receiver who took a step upfield, got hit and spilled the ball. This play had the biggest change in GWC of the gane, and the negative outcome was nothing to do with Tannehill. This is recorded officially as an interception. For argument’s sake if it was recorded as an 18 yard completion that became a forced fumble, then Tannehill’s passer rating goes up to about 110.

So the Carolina game is arguable either a failure of coaching or a failure of Tannehill. I will let better tape watchers draw their conclusions.

The Houston game, as far ad I am concerned showed Tannehill produced in a high volume game, with his rating being knocked down by factors beyond his control.

in addition Tannehill’s 3rd highest volume game was a 33 attempt game against Tampa, which was a win where he recorded a 109.8 rating.

"The Guy wrote:

The question is, "why does his performance diminish so much more than that of QBs with similarly high overall passer ratings in high-volume games."

You haven’t established that it does. You’ve spot checked a couple of QBs. You haven’t looked at Cousins, Garoppolo, Carr, Jackson, Stafford, Bridgewater, or Matt Ryan.

"The Guy wrote:

Interesting question. Stafford's career average passer rating in low-volume games is 97.2. In high-volume games it's 85.7.

For Luck those figures are 96.3 and 88.8.

Again for Tannehill those figures are 97.9 and 80.5. Once again his figure in high-volume games is a good bit lower (though less in these cases) than even quarterbacks people might consider average (as opposed to the earlier comparisons with Brees, Mahomes, etc.).

LOL. Now we get to the hair splitting.

"Pauly wrote:

Let’s look at these 2 datapoints
First you havethe 20-30 loss to Carolina. Tannehill had an 82.3 passer rating. Good to win about 40% of games if I’m doing the math right. Carolina got out to a 17 point lead in the first half and forced the Titans into pass first mode. Few QBs have good games once they get down 0-17. Game recaps said that the Titans gameplan did not establish Derek Henry and the run in the first half.

Second was the first game against Houston. Tannehill had a 92.2 rating, good for 50% chance of winning the game. However ne play in particular stands out and it was discussed at length at the time. Tannehilk threw an 18 yard pass to the 1 yard line which was caught by the receiver who took a step upfield, got hit and spilled the ball. This play had the biggest change in GWC of the gane, and the negative outcome was nothing to do with Tannehill. This is recorded officially as an interception. For argument’s sake if it was recorded as an 18 yard completion that became a forced fumble, then Tannehill’s passer rating goes up to about 110.

So the Carolina game is arguable either a failure of coaching or a failure of Tannehill. I will let better tape watchers draw their conclusions.

The Houston game, as far ad I am concerned showed Tannehill produced in a high volume game, with his rating being knocked down by factors beyond his control.

in addition Tannehill’s 3rd highest volume game was a 33 attempt game against Tampa, which was a win where he recorded a 109.8 rating.

But those two data points are high-volume games, where there is plenty of opportunity (given the high number of pass attempts) to overcome those arguably isolated incidents and perform well overall nonetheless. Why didn't that happen?

When you look at the best quarterbacks in the league over their careers as a whole, they apparently tend to do just that in those games, because their average passer ratings in such games are astronomically higher than Tannehill's. Even quarterbacks who are more likely to be considered just average (like Stafford and Luck for example; see a few posts above) have performed a good bit better than Tannehill in such games.

"Pauly wrote:

Let’s look at these 2 datapoints
First you havethe 20-30 loss to Carolina. Tannehill had an 82.3 passer rating. Good to win about 40% of games if I’m doing the math right. Carolina got out to a 17 point lead in the first half and forced the Titans into pass first mode. Few QBs have good games once they get down 0-17. Game recaps said that the Titans gameplan did not establish Derek Henry and the run in the first half.

Second was the first game against Houston. Tannehill had a 92.2 rating, good for 50% chance of winning the game. However ne play in particular stands out and it was discussed at length at the time. Tannehilk threw an 18 yard pass to the 1 yard line which was caught by the receiver who took a step upfield, got hit and spilled the ball. This play had the biggest change in GWC of the gane, and the negative outcome was nothing to do with Tannehill. This is recorded officially as an interception. For argument’s sake if it was recorded as an 18 yard completion that became a forced fumble, then Tannehill’s passer rating goes up to about 110.

So the Carolina game is arguable either a failure of coaching or a failure of Tannehill. I will let better tape watchers draw their conclusions.

The Houston game, as far ad I am concerned showed Tannehill produced in a high volume game, with his rating being knocked down by factors beyond his control.

in addition Tannehill’s 3rd highest volume game was a 33 attempt game against Tampa, which was a win where he recorded a 109.8 rating.

Well explained. Perfectly reasonable approach.

"The Guy wrote:

But those two data points are high-volume games, where there is plenty of opportunity (given the high number of pass attempts) to overcome those arguably isolated incidents and perform well overall nonetheless. Why didn't that happen?

When you look at the best quarterbacks in the league over their careers as a whole, they apparently tend to do just that in those games, because their average passer ratings in such games are astronomically higher than Tannehill's. Even quarterbacks who are more likely to be considered just average (like Stafford and Luck for example; see a few posts above) have performed a good bit better than Tannehill in such games.

Prove your statement. You cant just make that sort of assertion without proof.

A quick check of the stats shows that Tannehill has attempted a far greater % of his passes while trailing that the QBs he is being compared to. That is a function of the quality of the Dolphins’ teams he played on.

"The Guy wrote:

But those two data points are high-volume games, where there is plenty of opportunity (given the high number of pass attempts) to overcome those arguably isolated incidents and perform well overall nonetheless. Why didn't that happen?

When you look at the best quarterbacks in the league over their careers as a whole, they apparently tend to do just that in those games, because their average passer ratings in such games are astronomically higher than Tannehill's. Even quarterbacks who are more likely to be considered just average (like Stafford and Luck for example; see a few posts above) have performed a good bit better than Tannehill in such games.

LOL “astronomically higher”...... now hyperbole is the tactic....

"cbrad wrote:

You're reading the graph wrong. The x-axis is passer rating not passing attempts. That graph is saying that the better the QB is (more to the right on the x-axis = higher passer rating), the correlation between passing attempts and passer rating becomes more negative. In other words, the better the QB is the worse the decrement becomes with higher number of passing attempts.

What FinFaninBuffalo said in post #5042 however does make some sense:

I can't test that, but it's plausible that the better a QB is the more they're asked to do when behind, leading to bigger decrements.

Sorry for the confusion. I realize that the x-axis is passer ratings. I should have been more explicit.

I feel that since a bad to average QB makes more mistakes when under less defensive pressure then a good QB, then getting pressured doesn't make him a whole lot worse, just worse. Think of it as having a failure rate that increases in a linear manner.

I feel with an above average QB who doesn't make as many mistakes under less pressure, there is a breaking point where, when under pressure, he "tries to put the team on his back". This means he will take even more extreme risks than lower quality QB's, which makes his failure rate "jump off the deep end". Think of it as having a knee in a curve that gets steeper geometrically. A good example might be Tom Brady under strong pressure. I saw this in his last game of the season and the first game of the playoffs and suspect it would fit your data.

I hope this makes my thoughts a little clearer than in my original post.

Another couple QBs we haven't looked at here.

Drew Brees's career average passer rating in low-volume games: 103.4. In high-volume games: 94.9. Again Tannehill's career figures are 97.9 and 80.5. Again we see the relatively small difference between them in low-volume games, and the huge difference between them in high-volume games.

Lamar Jackson's figures in 2019: 111.7 and 87.7, similar to Tannehill's in 2019 (118.6 and 87.3).

"The Guy wrote:

Another couple QBs we haven't looked at here.

Drew Brees's career average passer rating in low-volume games: 103.4. In high-volume games: 94.9. Again Tannehill's career figures are 97.9 and 80.5. Again we see the relatively small difference between them in low-volume games, and the huge difference between them in high-volume games.

Lamar Jackson's figures in 2019: 111.7 and 87.7, similar to Tannehill's in 2019 (118.6 and 87.3).

Why do you act like the receivers factor into this at 0%? If a QB is throwing to average or poor receivers, I wouldn't expect his rating to climb the more he has to throw. If the QB is routinely playing in games where the team is behind, I wouldn't expect to see his rating climb when he has to throw a ton. Blowout wins where the QB is allowed to throw a ton is where you will ratings climb with more throws.

Frankly, your conclusions are always so confusing.

"The_Dark_Knight wrote:

I’d be curious to see how that ranks all time

OK.. as promised.. I finally have the program to adjust by era for December ratings, at least since 1978. So it turns out that Tannehill's 124.6 December rating ranks 5th all time since 1978. Here are the leaders with 2019-adjusted ratings in parenthesis.

1. Kurt Warner 2001 (146.02)
2. Drew Brees 2019 (137)
3. Drew Brees 2011 (130.71)
4. Chad Pennington 2002 (125.15)
5. Ryan Tannehill 2019 (124.6)
6. Brett Favre 1995 (124.36)
7. Dan Marino 1991 (123.46)

But man.. Brees!