Evaluating Tannehill's Value

Brady's initial contract was cheap, but he's been paid top dollar for the last 14 years.
Wilson was cheap for his first 3 years and has been paid like a top QB since.
Cousins was paid peanuts for the first 3 years, but is in line for top 5 money for last year, this year and next year.
The same will happen with Prescott.

Once a QB plays at NFL starter level or higher, he gets paid at that rate. The only way to beat the system is to draft a 3rd rounder or lower and have him play at probowl level from day one and then repeat that process every 3 years.

 

Maybe the answer to the question posed in this thread can be conditioned on what Tannehill does in 2017?

Moore was ranked #12 among all starting QB's in 2011 after starting 12 games and playing in 13. Tannehill was ranked #12 among all starting QB's in 2016 after starting 13 games but not finishing the 13th. VERY similar stat.

So I think there's a good argument that Moore might have been the better option in 2012 or 2013. But there's also a reason he's a career backup: no coach/GM thinks he can consistently be ranked high among starting QB's.

So if Tannehill makes it into the top 10 this year then I think drafting Tannehill was worth it even if waiting 6 years was frustrating. But if he doesn't? Well.. then I think you keep your options open because the QB position is VERY important. I mean there's a reason teams "reach" for starting QB's.

 

Looking at NFL playoff teams.
I have data on team passer rating made and allowed from 2002 to 2016.
The long term trend is for a passer rating of 87.97 taking 2016 as the endpoint.
Adjusting PR to 87.97 as the base for comparison since PR has been increading by nearly 0.75 points per year.
Offensive Standard deviation is 12.12
Defensive standard deviation is 9.23

Average NFL playoff teams:
Passer rating made: 97.9
standard deviation: 11.04
Passer rating allowed: 85.7
standard deviation: 8.17
Average record: 11.1 - 4.9

on the raw numbers
passer rating made: 91.4
passer rating allowed: 80.0
NFL average: 82.1

So on average NFL playoff teams have sightly better than average passing defenses but much better than average passing offenses.

 

Just thinking out loud but wouldn't it be better to compare the playoff teams vs the non-playoff teams instead of the entire league as a whole (which includes the playoff teams so the numbers are skewed higher)

I've posted before but the greatest correlation with winning is passer rating differential

 

For non playoff teams
adjusted to 2016 base (87.97 average)
Passer rating made 84.6
passer rating allowed 92.8

raw figures (82.1 average)
passer rating made: 79.0
passer rating allowed: 86.6

Average record 6.1 - 9.9

As expected the biggest difference is in passer rating made, not passer rating allowed

 

Both are informative.

From a purely statistical point of view the real utility of distributions is so that you can answer questions about the probability of different events occurring. I think for most questions you might find interesting combining the data gives you the distribution you're interested in.

For example you'd want to combine both playoff and non-playoff teams if you wanted to know the probability that a given passer rating will put it within the top 12 out of 32 teams in the league (in the upcoming season), or the probability that the passer rating will cross the threshold for making the playoffs given the actual NFL rules for choosing a playoff team (you need the overall distribution to do the simulations).

However, if someone asked the question whether a specified passer rating X is more likely to belong to a playoff or non-playoff team, then you'd want to separate the distributions.

So it depends on what kind of question you want to answer.

 

To look at the latter question and Ryan Tannehill.
Last years 93.5 was closer to an historical playoff team (4.5 points down) than a non playoff team (8.9 up).
Without rehashing all the arguments there are solid reasons to believe. his ceiling is a 100 PR a season passer. The frustrating thing is that he hasn't been able to do it for a full season yet.

Basically for the last 3 years Tannehill has performed in between "Playoff QB" level and "non playoff QB level" giving ammunition to both sides of the debate.

 

I love Moore, huge fan, yet I'm not about to argue that the two are equal....Tannehill has a much higher ceiling with a stronger arm, better accuracy, much more mobility, etc. I still think Moore gets the nod in leadership, controlling the pocket and reading the field post-snap, but Tannehill is the better QB.

Still though, how can year six be a "make or break" situation for Tannehill? I mean, if we win the Super Bowl and he has a 100+ rating on the year, then sure....we call him a superstar. But if he comes in anywhere from the mid 80's to mid 90's (which he has done every single year so far), what does that really prove? Or say he comes in as the #7 quarterback and we finish the season 8-8. We can create a million scenarios where the numbers say one thing, our eyes say something else and the coaches ultimately say, "I think he'll do better next year once we do x, y, and z."

So I guess my real question is- if we haven't cut him in five years, then what would he have to do in year six that would be a deal breaker? I can't think of a single scenario where that happens unless he's injured. The fact is that the longer we invest in him thinking that he's our franchise QB, the harder it becomes to admit that he's not. And if that's the case, then what's the point of debating this at all? I believe he's our franchise QB for his entire career regardless of how good/bad he plays in 2017 and beyond.

 

Ah.. it's a little more complicated than that. Can't just look at summary stats and figure this out.

What you have to do is to fit the best fitting normal distribution to the playoff and non-playoff data and see where those distributions have equal height (equal likelihood). So I did this just for the last 3 years because: 1) the average passer rating the last 3 years is about the same but before that it's not, and 2) sample size starts to become decent with 36 playoff teams and 60 non-playoff teams (12 is a bit small).

So here are the best fitting normal distributions:
https://ibb.co/dUKNFv

I used raw data with no adjustment (arguably there is none needed for the last 3 years). The mean for non-playoff is 85.68 with Std. Dev = 7.99; mean for playoff is 95.72 with Std. Dev = 9.81.

As you see those two curves cross at 95.2. So for the last 3 years any passer rating below 95.2 is more likely to come from a non-playoff team while any passer rating above 95.2 is more likely to come from a playoff team.

Thus, Tannehill hasn't produced at that level yet.

 

There are always exceptions to the rule.

Yes you could have Tannehill be top 10 in passer rating with it really being one of those cases where passer rating failed to capture how good a QB he was. Conversely, he might fall outside of the top 10 in passer rating yet have played well enough so that in reality (whatever that really is for QB's) he should have been ranked top 10 as a QB.

But over many games and years simple stats like passer rating tend to tell a better story than biased and incomplete human memory of eyeball analysis that for a particular play or even game might be far more accurate.

So looking at where Tannehill is ranked by passer rating is a decent way of seeing how he's progressing (or not), and I think if you judge him that way you'll find it seems "fair" (e.g. compare passer rating for his rookie year, or for his best year last year etc..)

You may be right we might not cut him unless he truly stinks. That would just go to show what Gase and our GM prioritize as well as how difficult it may be to find a good replacement. But you never know! I mean.. cap-wise we don't take a bit hit if we cut in after 2017, so it's hard to tell where the threshold is.

Either way, what I was stating is my opinion of what we should do, not a prediction of what Gase will do.

 

That's really cool.
For me the part that is frustrating is that he is between tbe peaks. If he was on the left side of the red curve, you get rid of him, no question. If he's on the right side of the blue curve you keep him, no question.
The longer he stays in the in between zone the more frustrating it becomes, and at some point you are going to have to cut bait.

 

Just because RT was injured doesn't mean he wasn't a playoff caliber QB. I realize that some don't want to give RT credit for the Dolphins making the playoffs last season, but he was a very big piece of the puzzle.

There were 6 other playoff QB's last season that had a passer rating right at 95 (Big Ben) or below, Osweiler-72, Smith-91, Eli-86, Stafford-93, Wilson-92.

 

A perfect example would be to look at RT's and MM's passer ratings. MM had a 105 and RT had a 93 so of course some want to tout that as Moore playing better or even being a better QB. However, MM had a lower comp%, higher INT%, and averaged about 50 fewer passing yards per game. The biggest stat that enabled MM to have a higher rating was his TD%. Which was an unbelievable, and unsustainable, 9.2. Matt Ryan had the next best at 7.1 to put that in comparison.

 

Yeah I don't know if you're understanding the graph.

Since the mean for the blue curve (playoff team passer ratings) is at 95.72, that means half of all playoff teams in the last 3 years had passer ratings below that. So it's no surprise you'll find many examples of passer ratings below 95.2.

Secondly, the curves are really telling you the odds of a passer rating coming from a playoff team. So the height of the red curve at 93.5 (Tannehill's rating last year) is 0.01933 while the height of the blue curve at 93.5 is 0.01486.

The ratio of those two numbers (blue over red.. so playoff over non-playoff) is 0.7688 meaning that the odds of a QB with a 93.5 passer rating being a playoff QB is 3:4. So if you have a 93.5 passer rating 3 out of 7 times you will be on a playoff team. That's what that means.

Also.. the graph tells you nothing about injury. Of course Tannehill was more important in getting us to the playoffs than Moore. The only influence Moore had in that graph above is that he raised team passer rating for the Dolphins slightly (hardly makes a difference over 96 teams).

 

Moore just didn't have enough passing attempts for passer rating to be compared directly like that. You generally want at least 150 passing attempts in a season before you directly compare ratings of two QB's. Moore had a total of 123 including his playoff game. So my argument against a direct Tannehill vs. Moore passer rating comparison is don't directly compare ratings without acknowledging sample size being an issue.

 

I understand the graph. I'm just not sure it proves, or disproves, much. Saying that a QB has a better chance of making the playoffs with a 95 passer rating sounds good, but in reality there were QB's last season with ratings in the 80's and 70's who made the playoffs. Conversely, there were QB's with 95+ who didn't make the playoffs. So my point is I just don't think this stat is a good indicator of who may or may not make it to the playoffs.

 

I agree. And if you go back to any of those debates you'll see that my position has always been MM doesn't have a big enough sample size to make any conclusions. And while I can say that he had some very good moments while starting, he had more bad.

 

The graph is probably the best indicator you'll find because it quantifies the odds of making the playoffs for any given passer rating. Not sure any other stat I've seen does that.

 

Five of the top 10 QB's, in regards to passer rating, didn't make the playoffs last season. All of them had a 95+ passer rating.

 

Yup.. look at the graph and you'll see that's true but the odds get smaller and smaller. You're just not understanding this.

 

lol...I assure you I'm understanding. I just don't think that looking ONLY at passer rating, over a large number of seasons, will get you better than a 50/50 chance of predicting a playoff team. You say that you've yet to see a stat that is a better indicator. Well, what about wins/losses? lol...Is that not a stat? I kid. I kid..

 

lol..

Yeah I said it does a better job of telling you whether you're a playoff team or not IF you're using passer rating. IF you use passer rating, then that's the best indicator I think you can actually come up with because that IS the historical data.

Of course you're right one shouldn't look only at passer rating, but that's a separate story. Can't dispute the utility of the graph by pointing out there were passer ratings above or below some threshold.

 

You're right, it doesn't prove anything, except that QB ratings over 95 have a better chance of coming from a playoff team, which is all he's saying.

Lets continue to look at last year, there were 12 starting QBs with a rating over 95, half of them were playoff QBs, there were at least 20 starting QBs under that number and less than a 3rd were playoff QBs.

Numbers here are standing up to age old reason, good QBs can take middling teams to the playoffs, when the most important player on your team is the best player on your team your chances of winning increase drastically.

 

There were actually 11 starting QB's last season with a passer rating of 95 or higher and 19 under 95. Out of those 11, 5 played in the playoffs. (To be fair, Carr was hurt and had a 96, so you could say 6 of those 11).

Out of the 19 with under a 95, 6 made the playoffs. (I'm giving credit to a hurt RT here)


Your "age old reason" makes no sense. Out of the 12 playoff teams, which team was "middling"?


The only teams that a person could possibly call "middling" (out of the playoff teams) are:

Dolphins---Good QB play
Houston--Bad QB play
Detroit--- Good Qb play
Giants---below average QB play
Raiders---Good QB play

So, 7 QB's had really good teams. (Symbiotic relationship)

3 "middling" teams had good QB play (The team was carried by their QB's for a large part)

2 "middling" teams had below average or worse QB play (So wouldn't this by reason mean that their teams were actually pretty good and yet they were held back by their QB's play?)


In conclusion, the "age old reason" shows that only 3 of the 12 playoff QB's "took" their middling teams to the playoffs.

 

I'm nowhere close to thinking he should be gone. I'm also not against keeping options open. I think Gase is a keeper. Seeing our owner's loyalty to Sparano and Philbin makes me think Gase is here as long as he wants to be. If Gase feels Ryan can't handle everything Gase wants to do, and he finds a diamond in the rough, something could happen. If Ryan becomes a QB where he's a 92-100 QB instead of a 88-93 QB, it's his job for as long as he wants.

 

This is true. Passer rating over time is a great stat, but over individual years there may be times when the rating doesn't capture the essence of the year.
It is heavily skewed towards completion percentage. It is fine most of the time, but every once in a while it gets thrown for a loop.

 

Just for completeness, I decided to try and answer the question danmarino posed of how well using passer rating to predict playoff vs. non-playoff compares to using win%.

As posted earlier this graph shows the likelihood any passer rating belongs to a playoff or non-playoff team:
https://ibb.co/dUKNFv

The comparable graph using wins in a season is this:
https://ibb.co/mJw6av

At any point on the x-axis, the point on the upper-most curve divided by the sum of the points on both the upper and lower most curves tells you how often (fraction of the time) you'll correctly predict whether that stat belongs to a playoff or non-playoff team.

So do that calculation for every point, weighted by how often that stat occurs in the league (that's easy here.. the sum of both the lower and upper curves tells you how often that stat occurs in the league because both together includes all teams), then sum the results and you have the average percent correct in predicting whether a team is a playoff or non-playoff team using that stat (passer rating vs. wins).

Results:
On average, using those curves I generated, you'll correctly predict whether a randomly chosen passer rating in a given season belongs to a playoff or non-playoff team 74.7% of the time, while using wins you'll correctly predict it 90.3% of the time. This is probably close to the best one can do: wins allows you to predict playoff vs. non-playoff ~90% of the time while passer rating might have a ceiling of ~75% correct. Those numbers seem high but keep in mind chance is at 50%, not 0%.

 

Thanks...and I stand corrected! lol.... I never would have guessed that it would be that high!