Ryan Tannehill

"The Guy wrote:

Some info about quarterback play in the playoffs between the 2010 and 2018 seasons (99 total games).

The era-adjusted (adjusted to 2018) average passer rating of winning teams in the playoffs is 103.7; standard deviation 24.7.

The era-adjusted average passer rating of losing teams in the playoffs is 82.6; standard deviation 26.2.

Using a two-sample t-test, these groups are significantly different at a probability level of 0.000000027.

So if we define a "winning quarterback performance in the playoffs" as something within a standard deviation of what typically wins in the playoffs, a quarterback must have a passer rating of at least 79.

A passer rating below 79 gives a quarterback only about a 16% chance of being part of a winning team in a playoff game. A passer rating below 54.3 gives a quarterback only about a 5% chance of being part of a winning team in a playoff game.

If we define a "winning quarterback performance in the playoffs" as something that gives a team at least a 50% chance of winning a playoff game, a quarterback must have a passer rating of at least 103.7.

Feedback from #cbrad would be appreciated if and when he has the time.

Yeah so I double checked the stats, and except for the issue we identified long ago with you using Excel and me using Matlab those are basically correct. I'm getting an average 2018-adjusted rating of 105.3 for the winner and 84.8 for the loser with standard deviations of 24.67 for the winner and 26.54 for the loser. Result of the t-test is a p-value of 0.000000067.

However, the result isn't surprising. Basically it's just confirming that QB's (in the playoffs or regular season) tend to have higher passer ratings when they win than when they lose lol, and that the difference is "real" and not consistent with the hypothesis that there is actually no real difference and that the difference you're observing is due to "random variation".

So the hypothesis test is concluding the obvious. What's useful however is knowing the actual distribution of adjusted ratings so that you can calculate the historical probability of winning in the playoffs given any rating.

"djphinfan wrote:

Question, do you think Tannehill could of had as good as year with this offense as Fitzpatrick had with the offense?

With what Miami had offensively? Yea, he would have done about the same, maybe a bit better than Fitzpatrick.

"djphinfan wrote:

Question, do you think Tannehill could of had as good as year with this offense as Fitzpatrick had with the offense?

Not unless he was coached and adapted to play differently than he does. You'd have to convince him to either jump off script on every play or just chuck it up for grabs. He isn't naturally inclined to do either.

Now the reverse question, could Fitz go as far with the Tennessee offense, including putting up the highest passer rating, leading the league in accuracy, etc, etc.

So, I concede, if you want a QB to lead a crap team to 5 wins, Fitz is your man. If you want to assemble a credible team and put it in the hands of a QB that is capable of challenging a defense all over the field with elite arm talent, then Tannehill is your guy.

"FinFaninBuffalo wrote:

Not unless he was coached and adapted to play differently than he does. You'd have to convince him to either jump off script on every play or just chuck it up for grabs. He isn't naturally inclined to do either.

Now the reverse question, could Fitz go as far with the Tennessee offense, including putting up the highest passer rating, leading the league in accuracy, etc, etc.

So, I concede, if you want a QB to lead a crap team to 5 wins, Fitz is your man. If you want to assemble a credible team and put it in the hands of a QB that is capable of challenging a defense all over the field with elite arm talent, then Tannehill is your guy.

I dont think our offense is necessarily built to maximize his strengths.

I dont think it's necessarily that Fitz does better with a crap team, but more where the talent on that crap team is and the philosophy.

"djphinfan wrote:

You forget all the context and parameters that came with those names.

For example, pre draft I said we should take Taylor in the 6th round and that he would be a steal for a qb taking in the 6th Round, when mostly everyone else mocked that he would even make it as a backup, he has proven that evaluation big time..played well over his draft stock, that is how I judge..

Kaepernick I was always a fan and amidst the majority saying the dude couldn’t play in this league I projected that he could, dude played his best game in the Super Bowl, should of win it, and has a nice run relatively speaking.

Fitzpatrick I don’t know what you mean, I never talked about him until he became a dolphin, but I sure like how he plays the position at his age,smart enough to know he has to use his entire skill set to beat defense, I did say for one game I would take him over Tannehill.

Alex smith Not sure why you are referencing him..

Anyways, you are not clear on any of these it seems, Mariota yes , geez, y’all make me admit that every week, it’s pretty funny, and interesting, how you and a few others won’t let it go..Tells me a lot.

This is a Tannehill thread so I’m not sure how it turned into pointing out things that some of you don’t like about me, so maybe we should stay on topic...

You’ve always knocked Tannehill for not using his athleticism the same way some other QBs do. I don’t believe a QB has to do that to be successful.

And you asked why I thought I believe you fall in love with mobile QBs. So I just listed a bunch of reasons why I believe that.

Those are all QBs that you’ve had a crush on at one time or another. And you propped them up while we had Tannehill (or after in Fitz’s case), while disregarding Tannehill as a waste of time.

So it was on topic.

"djphinfan wrote:

Anyways, you are not clear on any of these it seems, Mariota yes , geez, y’all make me admit that every week, it’s pretty funny, and interesting, how you and a few others won’t let it go..Tells me a lot.

Hey, let’s not act like I have some kind of personality flaw because I brought it up. It sucks but the Mariota thing isn’t going to die. I’m not saying I’m going to continue to throw it around. You like to toss out things like “put your money where your mouth is” or “at least I stick my neck out and make my evals”. So no doubt people are going to call you out on it.

But in this case I brought it up because it’s a DIRECT example or comparison of someone you loved and fawned over, to Tannehill who is someone you disregarded as not good enough.

I vehemently disagree with DJ about Tannehill and always have.

Having said that, his record is better than most other people's on here.

"Fin D wrote:

I vehemently disagree with DJ about Tannehill and always have.

Having said that, his record is better than most other people's on here.

I don’t believe it’s unfair to questions people’s evaluations or past record of doing so. Especially if they take themselves very seriously as an evaluator. Lucky for me I don’t make many proclamations.

(The players I can remember proclaiming I liked are Haskins, Derrick Henry and Devante Parker.)

"Mcduffie81 wrote:

I don’t believe it’s unfair to questions people’s evaluations or past record of doing so. Especially if they take themselves very seriously as an evaluator. Lucky for me I don’t make many proclamations.

(The players I can remember proclaiming I liked are Haskins, Derrick Henry and Devante Parker.)

You're not doing that really though.

You are harping on one of the few misses and ignoring the many more hits.

Deej has a type when it comes to QBs. But he's called numerous guys from Wilson, to Mahomes, to Watson, to CKap, etc.

"Mcduffie81 wrote:

I don’t believe it’s unfair to questions people’s evaluations or past record of doing so. Especially if they take themselves very seriously as an evaluator. Lucky for me I don’t make many proclamations.

(The players I can remember proclaiming I liked are Haskins, Derrick Henry and Devante Parker.)

I liked Vince Young back in the day I cant talk haha

I loved Parker though, I thought Tannehill had at least a good chance to succeed.

I was wrong about Wilson not making it. Hes become a top QB in the league.

I will say I wanted us to take Rodgers when he was in the draft. That might have ended up nice.

"Fin D wrote:

You're not doing that really though.

You are harping on one of the few misses and ignoring the many more hits.

Deej has a type when it comes to QBs. But he's called numerous guys from Wilson, to Mahomes, to Watson, to CKap, etc.

I’m not harping. I brought it up once to make a relative point in a Tannehill thread.

I know DJ has a type. That was part of my point.

"Mcduffie81 wrote:

I’m not harping. I brought it up once to make a relative point in a Tannehill thread.

I know DJ has a type. That was part of my point.

Harping wasn't a fair word to use on my part.

But my issue is, was is the Mariota miss more significant than all the hits relative to this discussion?

"Fin D wrote:

Harping wasn't a fair word to use on my part.

But my issue is, was is the Mariota miss more significant than all the hits relative to this discussion?

In this thread, hell yeah!!! Lol

"Mcduffie81 wrote:

In this thread, hell yeah!!! Lol

Can I take that as an admission that you understand you aren't being really fair, but are just trying to win an argument?

"cbrad wrote:

Yeah so I double checked the stats, and except for the issue we identified long ago with you using Excel and me using Matlab those are basically correct. I'm getting an average 2018-adjusted rating of 105.3 for the winner and 84.8 for the loser with standard deviations of 24.67 for the winner and 26.54 for the loser. Result of the t-test is a p-value of 0.000000067.

However, the result isn't surprising. Basically it's just confirming that QB's (in the playoffs or regular season) tend to have higher passer ratings when they win than when they lose lol, and that the difference is "real" and not consistent with the hypothesis that there is actually no real difference and that the difference you're observing is due to "random variation".

So the hypothesis test is concluding the obvious. What's useful however is knowing the actual distribution of adjusted ratings so that you can calculate the historical probability of winning in the playoffs given any rating.

Thanks cbrad -- appreciate the feedback as always. I thought the t-test might be necessary as a way of determining that the "winning" (versus the losing) distribution should be used to determine the probabilities of winning associated with various passer ratings, given that there is overlap between the winning and losing distributions. Is that accurate, or does it matter?

At any rate, what we've apparently found here is that a team doesn't get a 50% or better chance of winning from its passer rating in the playoffs unless it gets one of at least 105.3 (103.7 if the team is using Excel :smile:). That's a pretty staggering number at first glance in my opinion.

I'll do some research to determine how likely quarterbacks with various abilities are to produce a passer rating that high in the playoffs.

"Fin D wrote:

Can I take that as an admission that you understand you aren't being really fair, but are just trying to win an argument?

I explained why I think it’s fair and relative.

"The Guy wrote:

Thanks cbrad -- appreciate the feedback as always. I thought the t-test might be necessary as a way of determining that the "winning" (versus the losing) distribution should be used to determine the probabilities of winning associated with various passer ratings, given that there is overlap between the winning and losing distributions. Is that accurate, or does it matter?

At any rate, what we've apparently found here is that a team doesn't get a 50% or better chance of winning from its passer rating in the playoffs unless it gets one of at least 105.3 (103.7 if the team is using Excel :smile:). That's a pretty staggering number at first glance in my opinion.

I'll do some research to determine how likely quarterbacks with various abilities are to produce a passer rating that high in the playoffs.

You are completely ignoring the margin of victory. If a QB has a passer rating of 110 in a game that his team won by 35 points, he didn't "need" a passer rating of 110 to get the win. In addition, you have a chick and egg situation. Does the passer rating lead to the win or does playing with a lead (or conversely playing from behind) affect the passer rating. I propose that it sometimes one, sometimes the other, and sometimes both. Once again, you are using much too simplistic of an analysis. Just shows a lack of sophistication in both your understanding of stats and the game.

"The Guy wrote:

Thanks cbrad -- appreciate the feedback as always. I thought the t-test might be necessary as a way of determining that the "winning" (versus the losing) distribution should be used to determine the probabilities of winning associated with various passer ratings, given that there is overlap between the winning and losing distributions. Is that accurate, or does it matter?

There isn't a single way of determining the relation between win probability and passer rating. Each approach answers a different question.

If you want to know what the probability that a winning QB had passer rating X then you do what you just did: you look at the distribution of passer ratings for all wins and calculate based on that. There the t-test doesn't matter because you're conditioning on winning so you HAVE to use just the distribution of passer ratings when the QB was on the winning team.

If however you want to know the probability of winning GIVEN a certain passer rating in a single game, you need to do logistic regression. So on the x-axis is passer rating and y-axis goes from 0 to 1: every passer rating for a loss gets a 0 while every passer rating for a win gets a 1, and you fit a logistic function to it.

If however you want to know the probability of winning GIVEN a certain passer rating over multiple games (e.g., season ending passer rating), then you do linear regression with passer rating on the x-axis and win probability on the y-axis. That graph I already provided.

So for the probability of winning GIVEN a passer rating you always use both winning and losing distributions. It's only if you condition on winning or condition on losing that you use one or the other regardless of t-test. So the proper interpretation of the stats you have from the winning distribution is to say that IF the QB was on the winning team X% of the time the passer rating was at least Y, etc... But you can't say a passer rating of Y means the probability of winning is X% from that.

"cbrad wrote:

But you can't say a passer rating of Y means the probability of winning is X% from that.

I predict that he will still make the claim (perhaps reworded in some way), especially if the Titans manage to beat the Ravens. He NEEDS to try to diminish Tannehill's accomplishments.

"The_Dark_Knight wrote:

With what Miami had offensively? Yea, he would have done about the same, maybe a bit better than Fitzpatrick.

Hmm, with 0 run game, and the worst oline in football??

I don’t believe that to be true..and I don’t think that’s bashing the player either.. Ryan had better talent on offense during his tenure dont ya think? Not better coaching but can we say better talent?

Maybe the question being answered is, Ryan can play better if he has a dominant run game and good protection, but not so much if he doesn’t?

Is that fair?

"FinFaninBuffalo wrote:

Not unless he was coached and adapted to play differently than he does. You'd have to convince him to either jump off script on every play or just chuck it up for grabs. He isn't naturally inclined to do either.

Now the reverse question, could Fitz go as far with the Tennessee offense, including putting up the highest passer rating, leading the league in accuracy, etc, etc.

So, I concede, if you want a QB to lead a crap team to 5 wins, Fitz is your man. If you want to assemble a credible team and put it in the hands of a QB that is capable of challenging a defense all over the field with elite arm talent, then Tannehill is your guy.

Thank you for your honest answer, I think that’s a pretty interesting study no?

"FinFaninBuffalo wrote:

Not unless he was coached and adapted to play differently than he does. You'd have to convince him to either jump off script on every play or just chuck it up for grabs. He isn't naturally inclined to do either.

Now the reverse question, could Fitz go as far with the Tennessee offense, including putting up the highest passer rating, leading the league in accuracy, etc, etc.

So, I concede, if you want a QB to lead a crap team to 5 wins, Fitz is your man. If you want to assemble a credible team and put it in the hands of a QB that is capable of challenging a defense all over the field with elite arm talent, then Tannehill is your guy.

I would think it’s also fair to see with a better run game and much better protection Fitz’s numbers would improve.. not reach Tannehill numbers because Ryan does have innate accuracy, but better across the board..

"Mcduffie81 wrote:

You’ve always knocked Tannehill for not using his athleticism the same way some other QBs do. I don’t believe a QB has to do that to be successful.

And you asked why I thought I believe you fall in love with mobile QBs. So I just listed a bunch of reasons why I believe that.

Those are all QBs that you’ve had a crush on at one time or another. And you propped them up while we had Tannehill (or after in Fitz’s case), while disregarding Tannehill as a waste of time.

So it was on topic.

The bottom line is the record overall, so whoever I fall in love with, love being put into the context Relative to draft status and where they got picked, if they become successful or if I project unsuccessful, athleticism is just a trait to analyze.

My love for wilson, Mahomes, Murray, Mahomes, Watson, is different then my interest in Taylor going that late that year, Or Kaepernick and Kyler Murray being called a bust predraft, it’s all relative..

So no I don’t agree that I look into athleticism too much in a Qb, I look into who can get it done at a high level... some you miss on obviously but overall is how one should be judged..

"djphinfan wrote:

Hmm, with 0 run game, and the worst oline in football??

I don’t believe that to be true..and I don’t think that’s bashing the player either.. Ryan had better talent on offense during his tenure dont ya think? Not better coaching but can we say better talent?

Maybe the question being answered is, Ryan can play better if he has a dominant run game and good protection, but not so much if he doesn’t?

Is that fair?

Flip side, Henry wasn't dominating like this with Mariota...

"djphinfan wrote:

I would think it’s also fair to see with a better run game and much better protection Fitz’s numbers would improve.. not reach Tannehill numbers because Ryan does have innate accuracy, but better across the board..

I agree with this.