Ryan Tannehill is Typically a Loser in NFL "Quarterback Duels"

[Dolfanalyst]

As I'm sure folks can imagine, quarterback play is strongly correlated with winning in the NFL.

Even more strongly correlated with winning is the difference between one team's quarterback's performance and the other's.

For example, in the five years between 2007 and 2011, teams that won the QB rating battle (i.e., whose QB had a higher QB rating than that of the opponent) won the game 80% of the time.

Compare that to the yards per carry battle in the running game, for example, where the team on top won only 53% of the time, barely above the expected 50%.

Yesterday Joe Flacco carved up the Dolphins defense to the tune of 381 yards passing, 4 TDs, and 1 INT.

His QB rating was 119.2, well above the league average of 89.

Ryan Tannehill's QB rating was 63.1, far below the league average.

The QB differential between them was 56.1, which is very large, and highly predictive of the lopsided win (38-6) with which it was associated.

Obviously the game didn't have to go that way. Ryan Tannehill could've pulled off a Dan Marino-type game like he did against San Francisco, played very much like Flacco did yesterday, and the final score would've very likely been much, much closer.

In fact, I grew up watching Dan Marino play a host of games whose final scores were in the 40 to 40 range. In other words, just because the other team scored a high number of points, likely fueled in large part by its QB, didn't mean Dan Marino wasn't going to stay stride-for-stride with that QB and make the game highly competitive.

In other words, Dan Marino very often rose to the occasion in NFL "quarterback duels."

So let's take a look at how Ryan Tannehill has performed over his career when the opposing QB has played well, as Joe Flacco did yesterday.

I've chosen a cutoff for "playing well" as a QB rating of 110 or higher, since the top season QB rating in the league this year (Drew Brees) is around that number.

Since 2012 (his rookie season), Ryan Tannehill has been involved in 16 games in which the opposing QB has had a QB rating of 110 or higher.

Those opposing QBs have averaged QB ratings of 129.3 in those games. Ryan Tannehill's average QB rating in those games is 80.3.

The Dolphins are 3-13 in those games.

The average QB rating differential in those games is 48.9, with a standard deviation of 25.6.

To get a "control sample" if you will, let's take a look at what happens throughout the rest of the league when one of the two QBs in a game achieves a QB rating of 110 or higher.

There have been 84 such games in 2016. The average QB rating differential in those games is 32.7, with a standard deviation of 25.3.

That difference in QB rating differential (between Tannehill and the rest of the league) is statistically significant at a 98% level of probability. In other words, it's very unlikely due to chance.

So when the average NFL QB is in a "quarterback duel," he is typically outdueled by roughly 33 points. When Ryan Tannehill is in a quarterback duel, on the other hand, he is typically outdueled by roughly 49 points.

Now, what exactly does that 16-point difference in QB rating mean?

There have been 28 games in the league in 2016 in which one team's QB bettered the other's by a QB rating between 13 and 19 points (16 points give or take 3).

Those teams are 24-3-1 in those games, a win percentage of 86%. A roughly 16-point difference in QB rating equals a win probability of 86% in 2016.

What this all means in my opinion is that if the Dolphins are going to continue using Ryan Tannehill as their starting QB, they're going to have to invest a great deal of resources in their pass defense, such that they have the ability to limit the effectiveness of opposing QBs and minimize the number of "quarterback duels" Ryan Tannehill experiences.

Because when Ryan Tannehill experiences a quarterback duel, he performs significantly worse than the average QB in the league in that regard, which makes the Dolphins very likely to lose.

One can imagine what that might be like in the playoffs, where better QBs are typically faced, and the likelihood of a quarterback duel increases.

[Run Ricky Run]

Gase forget for a game that Tannehill is no better than a game managers.  Ajayi averaged 5yds a carry and only gets 12 touches?  While we force the ball to the douchebag Landry 14 targets and a straight up handoff (from the backfield for -1yds) for 78 total yards.  He is not a playmaker, stop pretending he is.  Gase gave up on the run game, like a typical idiot coach usually does.  First drive, Ajayi running good. He passes on 3rd and 2, and then attempts a fg on turf that everybody (well Miami had been, another idiotic coaching error) has been slipping on.  That is idiotic coaching.  How many bubble screens did we see get no yardage yesterday.  The last drive of the game, they completed 3 passes and didn't pick up the first down.  Does that type of offense scare anybody?

[fyo]

There's absolutely no indication that there is any causative effect at play here. None.

[Dolfanalyst]

There's absolutely no indication that there is any causative effect at play here. None.

To take that position you would have to believe that QB ratings, over decently-sized samples of play, are caused by something other than QBs' ability.  Do you really believe that?

[fyo]

To take that position you would have to believe that QB ratings, over decently-sized samples of play, are caused by something other than QBs' ability.  Do you really believe that?

What?

If a quarterback for team A can't win if when the opposing quarterback plays extremely well, it could just mean his defense sucks . Or it could just mean that your sample size is so small that what you're measuring is random noise. Or it could be that more are away. Or it could be that more are against actual quality teams and not just Sam Bradford's Vikings. Or it could mean any number of different things. It's entirely unclear what the underlying mechanics actually are in what you are measuring.

[Dolfanalyst]

What?

If a quarterback for team A can't win if when the opposing quarterback plays extremely well, it could just mean his defense sucks . Or it could just mean that your sample size is so small that what you're measuring is random noise. Or it could be that more are away. Or it could be that more are against actual quality teams and not just Sam Bradford's Vikings. Or it could mean any number of different things. It's entirely unclear what the underlying mechanics actually are in what you are measuring.

Here's some more info that may help:

The average QB with the higher QB rating in a quarterback duel league-wide is 123.4.  The one with the lower rating averages 90.7.

For the Dolphins since 2012 (Tannehill's rookie season), those numbers are 129.3 and 80.3.

Tannehill has been 10 points lower than the league average in such games.

[Dolfanalyst]

On second thought I appreciate the movement of this thread to this forum.  From now on when I post a negative Tannehill thread, I'll simply put it here.  There's nothing inherently wrong with this forum.  The people worth talking to are here, too.

[fyo]

I don't buy your concept of a "quarterback dual". Quarterbacks don't play each other and the argument that they directly influence each other is questionable at best. The statistical analysis is suspect due to, among other things, the lack of a homogeneous opponent pool, inherently different teams and defenses, and low sample size.

I looked up Aaron Rodgers' stats on this and he's 3-10 under the same circumstances, but with opponents playing at a much lower level. So what if *he* scored a higher rating. His team lost. Maybe he should have tried forcing a few more throws and gone for the win instead of maximizing his stats .

[Dolfanalyst]

The concept of a quarterback duel is sound because if one team gets an exceptionally good performance from its quarterback, it's very likely to win the game.  About the only thing that gives the opposing team a fighting chance is at least a very good performance by its own quarterback.  If its own quarterback lays an egg, there is an extremely low chance of that team's winning the game.

This is why the original post started off with the correlation between quarterback rating differential and winning.

[Dolfanalyst]

I looked up Aaron Rodgers' stats on this and he's 3-10 under the same circumstances, but with opponents playing at a much lower level. So what if *he* scored a higher rating. His team lost. Maybe he should have tried forcing a few more throws and gone for the win instead of maximizing his stats .

I didn't say there was any quarterback who is immune to it.  The loser of the quarterback duel is typically also the loser of the game.

Now, look up how many times Aaron Rodgers has been a winner in a game when he has posted a QB rating of 110 or higher.

[Dolfanalyst]

I didn't say there was any quarterback who is immune to it.  The loser of the quarterback duel is typically also the loser of the game.

Now, look up how many times Aaron Rodgers has been a winner in a game when he has posted a QB rating of 110 or higher.

I went ahead and did this.  In the time Tannehill's been in the league, Rodgers has had 24 games in which he's had a QB rating of 110 or higher.  He's 22-2 in those games.  In the two losses, the opposing QBs had ratings of 120.2 to his 131.8, and 129.5 to his 125.5.

The average QB rating for the opposing QB in the 22 of Rodgers's wins is 77.9.

Tannehill during that time span has had 10 such games (with ratings of 110 or higher).  Rodgers does it 45% of the time, Tannehill 13% of the time.

[Dolfanalyst]

One can imagine what that might be like in the playoffs, where better QBs are typically faced, and the likelihood of a quarterback duel increases.

And here's the data on that:

Since 2012 there have been 20 occasions in which a QB has posted a rating of 110 or higher in the playoffs.

In 12 games there was a QB who had a QB rating of 110 or higher, and the other team's QB had a rating below 110.

The teams with the QBs with ratings 110 or higher are 10-2 in those games.  In the two losses, the opposing QBs' ratings were 98.0 and 93.5.

Obviously if Ryan Tannehill posts his customary QB rating of 80 when facing a QB who posts a QB rating of 110 or higher, the Dolphins would've had 0% chance of winning any of those playoff games.

That's a major problem, folks.  Obviously you can't get to the playoffs and have 0% chance of winning, and expect to win a Super Bowl.

Again, the Dolphins would have to amass an all-world pass defense, and simply hope that it prevents the QBs on opposing playoff teams from playing that well.

Hell, just this year the Dolphins have faced four opposing QBs who have posted QB ratings 110 or better.

Their record in those games?

0-4.

Tannehill's average QB rating in those games?

70.9.  Horrendous.

[fyo]

I sincerely hope you know how shoddy your use of statistics is in the above.

On a different note, but related to your initial post -- and without in any way legitimizing your fantasy QB Battle Royal -- I looked at the numbers. It turns out that Miami (Tannehill) actually has a better record in games against 110+ passer rating quarterbacks than average (and median), despite those opposing quarterbacks having the second highest average passer rating of any team. Buffalo faced the highest average passer rating (129.3) and lost all 14 of those games. Miami faced an average passer rating of 129.0 and won, as you noted, 3 of 16 (0.188). The league average record (from 2012-now) when the opposing quarterbacks have 110+ passer rating is 0.143 with an average passer rating of 125.0. So the conclusion must be, that Tannehill "just wins" .

It's also interesting (or not) that the number of times teams have wound up allowing 110+ passer rating quarterbacks correlates very well with their own (low) average passer rating, with two notably outliers in Arizona (Palmer bad) and New Orleans (Brees good). Disregarding those outliers, the correlation coefficient is a substantial (in this context) -0.54, which is higher than between passer rating and winning in the league overall. Of course, this is almost certainly just happenstance, unless you believe quarterbacks that do well in faux quarterback duels scare the opponents so badly, they don't show up. The three highest average own quarterback ratings against 110+ passer ratings also have the three lowest number of such games. They also "just happen" to have played against three of the worst five average opposing passer ratings, which would seem to indicate that *defense* might actually be playing a role here. Just maybe.

Correlation is not causation.

(And your sample size is ridiculous.)

[Dolfanalyst]

I sincerely hope you know how shoddy your use of statistics is in the above.

On a different note, but related to your initial post -- and without in any way legitimizing your fantasy QB Battle Royal -- I looked at the numbers. It turns out that Miami (Tannehill) actually has a better record in games against 110+ passer rating quarterbacks than average (and median), despite those opposing quarterbacks having the second highest average passer rating of any team. Buffalo faced the highest average passer rating (129.3) and lost all 14 of those games. Miami faced an average passer rating of 129.0 and won, as you noted, 3 of 16 (0.188). The league average record (from 2012-now) when the opposing quarterbacks have 110+ passer rating is 0.143 with an average passer rating of 125.0. So the conclusion must be, that Tannehill "just wins" .

It's also interesting (or not) that the number of times teams have wound up allowing 110+ passer rating quarterbacks correlates very well with their own (low) average passer rating, with two notably outliers in Arizona (Palmer bad) and New Orleans (Brees good). Disregarding those outliers, the correlation coefficient is a substantial (in this context) -0.54, which is higher than between passer rating and winning in the league overall. Of course, this is almost certainly just happenstance, unless you believe quarterbacks that do well in faux quarterback duels scare the opponents so badly, they don't show up. The three highest average own quarterback ratings against 110+ passer ratings also have the three lowest number of such games. They also "just happen" to have played against three of the worst five average opposing passer ratings, which would seem to indicate that *defense* might actually be playing a role here. Just maybe.

Correlation is not causation.

(And your sample size is ridiculous.)

Of course defense plays a role.  And so do the quarterbacks.  Again, unless you really believe that QB ratings over fairly large sample sizes aren't indicative of QBs' ability.

[fyo]

Of course defense plays a role.  And so do the quarterbacks.  Again, unless you really believe that QB ratings over fairly large sample sizes aren't indicative of QBs' ability.

The sample size isn't large. Some teams on the list have a handful of these games. The stat you pimped in your last post had a 4 game sample size for the Dolphins.

Look, it's abundantly clear that the teams the Dolphins faced and the teams X other team faced in your made-up Quarterback Duel are not comparable. Just look at them and the teams they played. The teams themselves are also very different. For such differences to be even out, large sample sizes are needed and 5-6 games isn't remotely close to cutting it. Even worse, the top teams (those with the highest average "own passer rating") are also those with the (by far) fewest samples making them very suspect. They also happen to feature some killer defenses (Denver and Seattle), which I would argue is a much better explanation that your fictitious quarterback duel concept.

Even if EVERYTHING was accurate and there was such a thing as your Quarterback Duel, it's still completely irrelevant. The league average is a win record of 0.143 when facing 110+ passer rating quarterbacks since the 2012 season. Over the past 4.75 seasons, the Dolphins have had 16 such games, slightly over the league average of 15.25. That's just over 3 games a season (on average). The BEST win record for a team with at least 10 such games (to get even a bare minimum of samples, but still way too small to compare) is New England with 0.333. If your fantasy were true, Tanehill's "Loser" status costs the Dolphins less than half a game a season (0.6 for a league average team, but the Dolphins are a bit better than average against 110+ quarterbacks).

Let me repeat that: Even if your fantasy were true, elevating Tannehill's performance to among the best in the league would only result in an increase of 0.5 games a season. And that assumes quarterback is not just the biggest factor, but the ONLY factor, in winning games where opposing defenses have a 110+ rating.

As they say: Lies, damned lies, and statistics.