Ryan Tannehill

"resnor wrote:

Patriots...

I would officially disassociate myself with everything NFL.

"PhinFan1968 wrote:

I don't expect his play to decline. I believe he's helped partially by only having to have played 7.5 games compared to a full season...less wear and tear, etc. He and the Titans are currently #1 in passer rating, #1 in redzone TD percentage, #1 in YPA, #2 in completion percentage (behind only Drew Brees)...mostly against top 10 pass defenses.

Those are some great numbers for "just a guy."

I like that Tennessee is in full control over their own destiny with playing the Texans twice in a two-week span. They're a virtual lock at 10 wins, but even if they don't, 9 wins might get them in since there's very few competing for that last wildcard slot (Buffalo or NE is basically locked for the 1st wildcard at 9-3 and 10-2). So it's the 7-5 Steelers and 7-5 Tennessee with the 6-6 Raiders and Colts playing catch-up. The Steelers still have the Ravens and Bills left to play (plus Cardinals and Jets) so I don't think they'll get to 10 wins.

I broke that down so folks can see that Tennessee/Tannehill is in excellent shape if he can win in December...although that playoff berth will definitely have to be earned. A sweep of the Texans + 1 win gives them the division, or any three wins should give them the wildcard. But they still have to win AT LEAST 2 of the last 4 in any order.

"resnor wrote:

Patriots...

Belichek has certainly shown interest when RT went off in that 35-34 loss a few years ago (where Parker dropped the game-tying TD with nothing on the clock). I've sort of expected him to end up there all along and it's really going to ruin pro football for me.

The good news is that at the level he's playing, there's no way Tennessee lets him walk UNLESS he completely self-destructs and doesn't make the playoffs. And hey, we know it's entirely possible...but I'm crossing my fingers for the time being and hoping he stays on his current trajectory.

"KeyFin wrote:

Belichek has certainly shown interest when RT went off in that 35-34 loss a few years ago (where Parker dropped the game-tying TD with nothing on the clock). I've sort of expected him to end up there all along and it's really going to ruin pro football for me.

The good news is that at the level he's playing, there's no way Tennessee lets him walk UNLESS he completely self-destructs and doesn't make the playoffs. And hey, we know it's entirely possible...but I'm crossing my fingers for the time being and hoping he stays on his current trajectory.

Actually there is a crazy theory out there that Titans might let Tannehill go if they think they can get Brady... I have had this sense of doom that this might come to pass ever since we decided to let go of Tannehill.

"PhinFan1968 wrote:

I would officially disassociate myself with everything NFL.

I've always thought that NE was a place where Tannehill would flourish.

"resnor wrote:

I've always thought that NE was a place where Tannehill would flourish.

Oh, I think he would have an incredible career there and probably end up breaking records. I just don't want to be a Pats fan so I'm hoping it doesn't happen.

"PhinFan1968 wrote:

Ya anything's possible...but the one thing that keeps me from expecting it to decline is the quality of pass defenses he has faced and will face through the end of the season...he's already gone through 4 top 10 pass defenses and done great against them.

You don't say how you're measuring pass defense, but I would encourage you to measure it with passer rating surrendered because of its correlation with winning.

Here are the z-scores of the teams Tannehill has played against in terms of passer rating surrendered -- the higher the number the worse the pass defense as compared to the current league average:

Denver: -0.03
Chargers: 0.54
Tampa Bay: 0.41
Carolina: -0.73
Kansas City: -0.47
Jacksonville: 0.07
Indianapolis: 0.55

Two of those teams are top-10 in the league in passer rating surrendered, Carolina (6th) and Kansas City (8th). Tannehill's own passer ratings against them were 82.3 and 133.9, respectively.

"The Guy wrote:

You don't say how you're measuring pass defense, but I would encourage you to measure it with passer rating surrendered because of its correlation with winning.

Here are the z-scores of the teams Tannehill has played against in terms of passer rating surrendered -- the higher the number the worse the pass defense as compared to the current league average:

Denver: -0.03
Chargers: 0.54
Tampa Bay: 0.41
Carolina: -0.73
Kansas City: -0.47
Jacksonville: 0.07
Indianapolis: 0.55

Two of those teams are top-10 in the league in passer rating surrendered, Carolina (6th) and Kansas City (8th). Tannehill's own passer ratings against them were 82.3 and 133.9, respectively.

Why not just go by points allowed? I mean, that's what wins/loses games.

"cuchulainn wrote:

Amusing that Tannehill has one less sack (24) than Mariota (25) after both started six games.

With almost 50 sacks, Titans fans are far more critical of their OLmen than many Dolphins's fans. As a team, they're on pace to have ~65 sacks on the season.

Lewan in particular draws a ton of criticism and ridicule. Many want Conklin gone/replaced. The Titans are actually at the bottom of the league with the Dolphins in pass protection rating.

What I'm leaning toward personally to explain Tannehill's success is a synergy between Tannehill and the team such that Tannehill has to play very well to remain the starter, more so than at any other point in his career, while the team (if the above rumors are true) wanted Tannehill to be the starter out of camp and finally got their wish.

Put those two variables together and you have a very determined individual player and a team determined to play well enough to keep him the starter. That's a very strong team dynamic that elevates the play of everyone involved.

"resnor wrote:

Why not just go by points allowed? I mean, that's what wins/loses games.

Because that doesn't restrict the measurement to pass defense as well as does passer rating surrendered.

"The Guy wrote:

Here are the z-scores of the teams Tannehill has played against in terms of passer rating surrendered -- the higher the number the worse the pass defense as compared to the current league average:

Finally someone else calculating z-scores!! It's about time because raw ratings are just misleading, especially across years.

Only thing I'll add is that I always use positive z-scores for "above average ability" and negative z-scores for "below average ability" whether it's offense or defense because that makes interpreting things easier (without extra disclaimers).

"The Guy wrote:

What I'm leaning toward personally to explain Tannehill's success is a synergy between Tannehill and the team such that Tannehill has to play very well to remain the starter, more so than at any other point in his career, while the team (if the above rumors are true) wanted Tannehill to be the starter out of camp and finally got their wish.

Put those two variables together and you have a very determined individual player and a team determined to play well enough to keep him the starter. That's a very strong team dynamic that elevates the play of everyone involved.

Very plausible...and thanks for the z-score stuff...very interesting.

"The Guy wrote:

You don't say how you're measuring pass defense, but I would encourage you to measure it with passer rating surrendered because of its correlation with winning.

Here are the z-scores of the teams Tannehill has played against in terms of passer rating surrendered -- the higher the number the worse the pass defense as compared to the current league average:

Denver: -0.03
Chargers: 0.54
Tampa Bay: 0.41
Carolina: -0.73
Kansas City: -0.47
Jacksonville: 0.07
Indianapolis: 0.55

Two of those teams are top-10 in the league in passer rating surrendered, Carolina (6th) and Kansas City (8th). Tannehill's own passer ratings against them were 82.3 and 133.9, respectively.

So how do Oakland, New Orleans and Houston show for that measurement?

Am I the only one that finds it amusing that so many posters who are glad Tannehill is no longer a Dolphin are upset by his doing well with the Titans?

If you didn't like Tannehill to begin with, why are you even posting in this thread?

"PhinFan1968 wrote:

So how do Oakland, New Orleans and Houston show for that measurement?

Current z-scores for pass defense (lol.. sorry about the size.. anyway what's amazing is NE):

"PhinFan1968 wrote:

So how do Oakland, New Orleans and Houston show for that measurement?

z-scores (transformed so that the lower the score the worse the performance, like cbrad said above):

Oakland: -1.16 (30th in the league)
New Orleans: 0.43 (10th in the league)
Houston: -0.82 (25th in the league)

So he has an easy go of it from here on out.

If and when he faces New England in the playoffs -- z-score of 3.27 (1st in the league by a country mile -- next-highest team is at 1.67) -- we'll see what happens. The passer rating New England has surrendered this year is 57.

"cbrad wrote:

Current z-scores for pass defense (lol.. sorry about the size.. anyway what's amazing is NE):

We must've been typing that at the same time, but for some reason we're getting slightly different results.

"The Guy wrote:

We must've been typing that at the same time, but for some reason we're getting slightly different results.

Interesting.. might be a rounding error? Anyway, if you want to do this manually (I have a program that scrapes the data automatically and does the calculations) all I'm doing is literally copying the passer rating column here:
https://www.pro-football-reference.com/years/2019/opp.htm#all_passing

Then I calculate both the mean based on that column and the standard deviation and calculate the z-score. You can't trust pro-football-reference's own calculations because I think they round off before averaging. For example, they list the average passer rating allowed as 91.1 but if you actually do the calculations it's 90.9969. Anyway, not sure where the small discrepancy is coming from but I'm doing all the calculations from that column of passer ratings in that link.

"The Guy wrote:

z-scores (transformed so that the lower the score the worse the performance, like cbrad said above):

Oakland: -1.16 (30th in the league)
New Orleans: 0.43 (10th in the league)
Houston: -0.82 (25th in the league)

So he has an easy go of it from here on out.

If and when he faces New England in the playoffs -- z-score of 3.27 (1st in the league by a country mile -- next-highest team is at 1.67) -- we'll see what happens. The passer rating New England has surrendered this year is 57.

Hell, that top 10 list is a who's-who of playoff teams...seems to be a correlation maybe?

Minus the Saints, ya, should be smooth sailing this last month...of course, they have to play the games.

Could you please fill me in on how those scores are determined? And what site did you get the "passer rating surrendered" info from?

"cbrad wrote:

Interesting.. might be a rounding error? Anyway, if you want to do this manually (I have a program that scrapes the data automatically and does the calculations) all I'm doing is literally copying the passer rating column here:
https://www.pro-football-reference.com/years/2019/opp.htm#all_passing

Then I calculate both the mean based on that column and the standard deviation and calculate the z-score. You can't trust pro-football-reference's own calculations because I think they round off before averaging. For example, they list the average passer rating allowed as 91.1 but if you actually do the calculations it's 90.9969. Anyway, not sure where the small discrepancy is coming from but I'm doing all the calculations from that column of passer ratings in that link.

That's weird because I'm taking it from the same place and getting an average of 90.99375.

"PhinFan1968 wrote:

Hell, that top 10 list is a who's-who of playoff teams...seems to be a correlation maybe?

Correlation between offensive passer rating and win% across NFL history is 0.633 meaning that 0.633^2 = 40.07% of the variation in win% is explained by passer rating. For defensive passer rating the correlation is -0.593 so 35.16% of the variance in win% is explained by it. For passer rating differential it goes up to 0.7951 correlation and 63.22% variance explained. So it's a very important stat.

"PhinFan1968 wrote:

Could you please fill me in on how those scores are determined? And what site did you get the "passer rating surrendered" info from?

Data are from pro-football-reference:
https://www.pro-football-reference.com/years/2019/index.htm#all_passing
https://www.pro-football-reference.com/years/2019/opp.htm#all_passing

To calculate a z-score take a given rating, subtract the mean from it, then divide by the standard deviation. That's it.

So for example the Dolphins have a passer rating surrendered of 105.3, the mean is 90.9969 (see my post above) and the standard deviation is 10.5444. So you get (105.3-90.9969)/10.5444 = 1.3565, and since it's defense we just multiply by -1 to put everything on an "ability scale". So the z-score for Miami's pass defense is -1.3565.

"PhinFan1968 wrote:

Hell, that top 10 list is a who's-who of playoff teams...seems to be a correlation maybe?

Minus the Saints, ya, should be smooth sailing this last month...of course, they have to play the games.

Could you please fill me in on how those scores are determined? And what site did you get the "passer rating surrendered" info from?

Definitely a correlation. And if you throw in offensive passer rating and use passer rating differential (offensive passer rating minus passer rating surrendered), and order teams from best to worst, you'll also have a very good approximation of the ordering of the records (win percentages) of all the teams in the league.

"The Guy wrote:

That's weird because I'm taking it from the same place and getting an average of 90.99375.

That's weird. I just manually checked this in Matlab (usually I use a scraper in R) and it comes out the same. I deliberately formatted so that it can output up to 16 significant digits but PFR's data is only out to 3 past the decimal point, so we should be getting the same thing.

What program are you using? Just curious.

"cbrad wrote:

Correlation between offensive passer rating and win% across NFL history is 0.633 meaning that 0.633^2 = 40.07% of the variation in win% is explained by passer rating. For defensive passer rating the correlation is -0.593 so 35.16% of the variance in win% is explained by it. For passer rating differential it goes up to 0.7951 correlation and 63.22% variance explained. So it's a very important stat.

Data are from pro-football-reference:
https://www.pro-football-reference.com/years/2019/index.htm#all_passing
https://www.pro-football-reference.com/years/2019/opp.htm#all_passing

To calculate a z-score take a given rating, subtract the mean from it, then divide by the standard deviation. That's it.

So for example the Dolphins have a passer rating surrendered of 105.3, the mean is 90.9969 (see my post above) and the standard deviation is 10.5444. So you get (105.3-90.9969)/10.5444 = 1.3565, and since it's defense we just multiply by -1 to put everything on an "ability scale". So the z-score for Miami's pass defense is -1.3565.

I took stats TOO long ago, and didn't really do well / pay much attention...how do you get standard deviation again? In simple terms that I can understand lol?

Edit: Oh wait...that's a calculation in Excel and other spreadsheets right? Comes built in?

"PhinFan1968 wrote:

I took stats TOO long ago, and didn't really do well / pay much attention...how do you get standard deviation again? In simple terms that I can understand lol?

Edit: Oh wait...that's a calculation in Excel and other spreadsheets right? Comes built in?

In Excel it looks like the command is =STDEV(). I say "looks like" since I googled it. I almost never use Excel.. way too cumbersome and can't do more advanced stuff. Here's a link that takes you step by step through the calculations of standard deviation:
https://www.khanacademy.org/math/probability/data-distributions-a1/summarizing-spread-distributions/a/calculating-standard-deviation-step-by-step

And once you understand that, note that for samples the standard deviation actually divides by N-1 not N as in that link (technical reason). And normally you'll see the formula with parentheses instead of absolute values (same result).