What a difference a line makes.

[Dolfanalyst]

I'm beginning to question if you watch football or just crunch numbers based on a statement like this. I don't buy it for a minute that lineman have greater chances of compensating for another's play. They should already be engaged in their own assignments, how can they cover two? If anyone is expected to compensate for a lineman's mistake it is a running back or maybe a tight end.

What I mean is that, when you're evaluating the play of the line as a whole, there is likely to be variation in the abilities of the linemen, such that functioning of the line as a whole is enhanced by the better players on it, thus compensating for the poorer players.

In other words, nobody has five Dallas Thomases.  On any team, there is likely to be a Dallas Thomas type (the worst guy), as well as a Richmond Webb type (the best guy, who is much better than the worst guy).  So the line as a whole is unlikely to be either tremendously good or tremendously poor.

Salary cap and parity also play into that, whereby teams are unlikely to have neither the resources to have a tremendously good line, nor the lack of resources to have a tremendously poor line.

Again, the idea is that I suspect there is much smaller variation among the lines across the league than many people seem to believe.

And if that's true, and given the weak correlation between offensive line and QB play, one would expect only a meager improvement in Ryan Tannehill's performance if the Dolphins had even the best offensive line in the league, because theoretically there isn't much room between their current line and the best one.

Really this whole "offensive line" thing I suspect has gotten even the team off into a wild goose chase, where they've sunken too many resources into the line, at the expense of talent elsewhere, thinking it would improve Tannehill's performance.

You could easily end up in a situation here where lots of limited resources are shoved into the line, the rest of the team is therefore weak, and the improvement in the line generated by those resources creates only a meager improvement in Tannehill's performance.

[fyo]

But in essence what you're saying here is that, in terms of the possibility that Tannehill's performance is being moderated by surrounding variables, he is experiencing a rare event.  And by definition a rare event is an unlikely event.

Now you are just being silly. On both accounts, by the way. A rare event is not an unlikely event if the history is included, which is the case here. Rolling 10 6'es in a row with a die is a rare event, but if you've already rolled 9, the event is not unlikely.

However, in the case of quarterbacks and careers, what I'm saying is that ALL quarterbacks are in fairly unique positions and, given the extremely low population count, trying to derive much of anything with any reasonable statistical significance is folly.

I actually went back and looked up some stats... there are only 95 quarterbacks with 60+ career starts (less than 4  seasons) going all the way back to the 80s! So this dataset would include Warren Moon, Boomer Esiason, and Dave Krieg.

Out of curiosity, I did manage to find a number of quarterbacks who track Tannehill fairly well when looking at stats like completion %, sack rate, and passer rating. In every case, of course, I can point to a specific (and different) likely cause of a dip or spike in performance as measured by those metrics.

However, if you just want an example of a quarterback among those 95 that has started out truly bad and wound up delivering some pretty solid numbers Alex Smith springs to mind. I found a few others as well. One common trend among them was that they all started to perform significantly better around year 5 or 6 (of course there's a huge selection bias at work here, since I was looking for players who wound up good, but started out bad or average). A player like Drew Brees also had some really rough years in San Diego before some great numbers in New Orleans.

[Dolfanalyst]

Now you are just being silly. On both accounts, by the way. A rare event is not an unlikely event if the history is included, which is the case here. Rolling 10 6'es in a row with a die is a rare event, but if you've already rolled 9, the event is not unlikely.

However, in the case of quarterbacks and careers, what I'm saying is that ALL quarterbacks are in fairly unique positions and, given the extremely low population count, trying to derive much of anything with any reasonable statistical significance is folly.

I actually went back and looked up some stats... there are only 95 quarterbacks with 60+ career starts (less than 4  seasons) going all the way back to the 80s! So this dataset would include Warren Moon, Boomer Esiason, and Dave Krieg.

Out of curiosity, I did manage to find a number of quarterbacks who track Tannehill fairly well when looking at stats like completion %, sack rate, and passer rating. In every case, of course, I can point to a specific (and different) likely cause of a dip or spike in performance as measured by those metrics.

However, if you just want an example of a quarterback among those 95 that has started out truly bad and wound up delivering some pretty solid numbers Alex Smith springs to mind. I found a few others as well. One common trend among them was that they all started to perform significantly better around year 5 or 6 (of course there's a huge selection bias at work here, since I was looking for players who wound up good, but started out bad or average). A player like Drew Brees also had some really rough years in San Diego before some great numbers in New Orleans.

It's interesting you mention Alex Smith, because I believe he's Tannehill's ceiling, personally, and Tannehill should be used the way Reid uses Smith if you want the most out of him.  The Pittsburgh game was a good example of that, where running and passing were balanced, the team rode the back of the running game, and Tannehill wasn't asked to take many shots down the field.  His good YPA that game (which again predicts winning) was driven mostly by his very good completion percentage (75%) and not his aggressiveness downfield.  Notice he had no INTs, as well, in a season in which he'd had an uncharacteristic percentage of them previously.

Drew Brees on the other hand had a great third year as a starter, after starting 27 of the 32 games the previous two seasons and zero games his rookie year.  With Tannehill we're working on four years of starting experience, and he has yet to show anything close to that level of play over a season.

[Spider-Dan]

Let's try to stay on the topic of the line, please.  If this becomes another thread entirely about Tannehill, it will be locked.

[Spider-Dan]

I would suggest you give that a try, via survey, and see if you get anywhere near the consensus you would if you were surveying people's impressions of QBs.

You have repeatedly emphasized that you are about statistical analysis, not subjective opinions.
Consensus of peoples' impressions (be it of an OL or a QB) are not "statistical analysis."  They are subjective opinions, and gathering more of them does not turn them into facts.

I have no problem with subjective opinions; everyone has their own impression of the various players on the team.  But don't try to disguise your personal distaste for specific players as objective statistical analysis, especially when you happily discard similar evaluations when inconvenient.

The same organizations that created metrics like DVOA have consistently rated the Dolphins' OL as poor (not "average to non-significantly below average") over Tannehill's tenure.  If you want to claim that the OL doesn't matter (or barely matters) in how the rest of the offense plays, that is your prerogative; I doubt many people will agree.  But you cannot, with any credibility, simply deny the poor evaluations of Miami's OL because there are too many outside factors and then cheerfully accept poor evaluations of other players you don't care for.

[Dolfanalyst]

You have repeatedly emphasized that you are about statistical analysis, not subjective opinions.
Consensus of peoples' impressions (be it of an OL or a QB) are not "statistical analysis."  They are subjective opinions, and gathering more of them does not turn them into facts.

I have no problem with subjective opinions; everyone has their own impression of the various players on the team.  But don't try to disguise your personal distaste for specific players as objective statistical analysis, especially when you happily discard similar evaluations when inconvenient.

The same organizations that created metrics like DVOA have consistently rated the Dolphins' OL as poor (not "average to non-significantly below average") over Tannehill's tenure.  If you want to claim that the OL doesn't matter (or barely matters) in how the rest of the offense plays, that is your prerogative; I doubt many people will agree.  But you cannot, with any credibility, simply deny the poor evaluations of Miami's OL because there are too many outside factors and then cheerfully accept poor evaluations of other players you don't care for.

First, you don't have any idea about how much "cheer" I have.

Second, you don't have any idea how much "distaste" I have for anything, or which players I "don't care for."

To pass yourself off has having such knowledge about me is arrogant, condescending, and rude.  Stop.

Third, Football Outsiders' measure of offensive line play has a 0.28 correlation with QBs' DVOA across the league.

In other words, 92% of the variation in QBs' DVOA is not associated with Football Outsiders' measure of offensive line play.  If you gave the Dolphins the very best offensive line play in the league as measured by Football Outsiders, then because of that weak correlation, you should expect a correspondingly meager improvement in Tannehill's DVOA.

So, be careful when you make these "adjustments" based on other areas of the team (e.g., the offensive line).  Any objective "adjustment" hinges on the strength of the relationship (correlation) between the two areas of the team.

Finally, here's a great treatise on how measurements can begin with subjective impressions and ultimately result in objective measures:

Even when a test is constructed on the basis of a specific criterion, it may ultimately be judged to have greater construct validity than the criterion. We start with a vague concept which we associate with certain observations. We then discover empirically that these observations covary with some other observation which possesses greater reliability or is more intimately correlated with relevant experimental changes than is the original measure, or both.

For example, the notion of temperature arises because some objects feel hotter to the touch than others. The expansion of a mercury column does not have face validity as an index of hotness. But it turns out that (a) there is a statistical relation between expansion and sensed temperature; (b) observers employ the mercury method with good interobserver agreement; (c) the regularity of observed relations is increased by using the thermometer (e.g., melting points of samples of the same material vary little on the thermometer; we obtain nearly linear relations between mercury measures and pressure of a gas). Finally, (d) a theoretical structure involving unobservable microevents -- the kinetic theory -- is worked out which explains the relation of mercury expansion to heat.

This whole process of conceptual enrichment begins with what in retrospect we see as an extremely fallible "criterion" -- the human temperature sense. That original criterion has now been relegated to a peripheral position. We have lifted ourselves by our bootstraps, but in a legitimate and fruitful way.

http://psychclassics.yorku.ca/Cronbach/construct.htm

Again, your validation of a measure of offensive line play would likely fail at "(b)" above, in that there would be poor interobserver agreement (i.e., consensus) about the quality of the offensive lines throughout the league.  I suspect that wouldn't happen for QBs, however.

[CF DolFan]

analyst ... Spin it any way you want. Whenever Tannehill has protection he does very well. Give him a running game too and he almost always wins. I'll take that over any spin doctoring statistic. 

[Dolfanalyst]

analyst ... Spin it any way you want. Whenever Tannehill has protection he does very well. Give him a running game too and he almost always wins. I'll take that over any spin doctoring statistic. 

You're the one doing the spinning, and it's based on statements that aren't true.

Tannehill's performance correlates strongly neither with sacks, nor with running game variables.

He's played well in a good number of games in which he's sacked comparatively more, and poorly in a good number of games in which he's sacked comparatively less.

Likewise, he's played well in a good number of games in which the running game performs comparatively poorly, and poorly in a good number of games in which the running game performs comparatively well.

These correlations that people believe exist do not exist.  If your position is based on the belief that a weak correlation in reality is actually a strong one, then you're the one doing the spinning.

[Spider-Dan]

Third, Football Outsiders' measure of offensive line play has a 0.28 correlation with QBs' DVOA across the league.

This is exactly what I mean.

You cannot hold up Football Outsiders' DVOA metric as golden because you think it supports your desired conclusion, and then dismiss Football Outsiders' offensive line ratings because they do not support your desired conclusion.  They are both subjective evaluations.  We've already discussed this.

And while we're on the subject of being arrogant, condescending, and rude: please don't act like we are unaware that you have been banned from several other Dolphins message boards specifically because of your single-minded devotion to bashing Tannehill at every turn.  I am aware of your history, and I'm still willing to have discussions about the QB when they are (at least tangentially) relevant, but I'm not willing to pretend that you are Just Making An Objective Analysis.

Any reasonably unbiased observer would either accept FootballOutsiders' metrics or dismiss them.  You simply dismiss the ones that say that maybe the problem is the line, not the QB... and not because you think they are inaccurate, but because you think there are too many variables to Truly Know The Answer, or something.  It's not even a credible objection.

[Dolfanalyst]

This is exactly what I mean.

You cannot hold up Football Outsiders' DVOA metric as golden because you think it supports your desired conclusion, and then dismiss Football Outsiders' offensive line ratings because they do not support your desired conclusion.  They are both subjective evaluations.  We've already discussed this.

In this case I'm not dismissing their offensive line statistic for the sake of argument.  I'm stating merely that it correlates weakly with QBs' DVOA, and so one should expect a meager improvement in Tannehill's DVOA even with the best offensive line as measured by Football Outsiders.

So, whether you believe their measure of offensive line play to be valid or not, realize that it doesn't do Tannehill any favors.

And while we're on the subject of being arrogant, condescending, and rude: please don't act like we are unaware that you have been banned from several other Dolphins message boards specifically because of your single-minded devotion to bashing Tannehill at every turn.  I am aware of your history, and I'm still willing to have discussions about the QB when they are (at least tangentially) relevant, but I'm not willing to pretend that you are Just Making An Objective Analysis.

Yet, who is the one being arrogant, condescending, and rude in this discussion (and others here)?  Me, or you?

[Dolfanalyst]

More regarding the difference a line makes:

http://www.footballoutsiders.com/stat-analysis/2016/quarterbacks-and-pressure-2015

The following table shows all 37 quarterbacks with at least 200 pass plays during the regular season. Scrambles and defensive pass interference penalties are included, but aborted snaps are excluded. The quarterbacks are sorted by DVOA difference, from largest difference to smallest. Sacks marked as "coverage sack" and scrambles marked as either "coverage scramble" or "hole opens up" are not counted as pressure plays. Scrambles go into DVOA as runs instead of passes, so the quarterbacks' statistics here will look different than their passing DVOA numbers.

[Dolfanalyst]

Based on the data at the link in the post above this one, what I said here is confirmed:

Again, the idea is that I suspect there is much smaller variation among the lines across the league than many people seem to believe.

And if that's true, and given the weak correlation between offensive line and QB play, one would expect only a meager improvement in Ryan Tannehill's performance if the Dolphins had even the best offensive line in the league, because theoretically there isn't much room between their current line and the best one.

The average line in the the league surrenders QB pressure on roughly 26% of its passing plays.

The average line deviates 4.6% from that figure.

The skewness of that distribution (0.28) isn't non-normal.

The negative kurtosis of the distribution (-0.66) indicates very light tails in the distribution.  In other words, most of the data is clustered around the average.

The frequency of pressure the Dolphins surrendered (28.1%) wasn't a standard deviation higher than the league average.  In other words, for all intents and purposes it was average.

So, the perception that the Dolphins' offensive line surrenders exceptionally frequent pressure is erroneous.

[fyo]

Again, your analysis completely ignores what actually happens on the field. If a team has a weak line, what do they do? They find other ways to compensate, at least to some extent, for it, even though those actions will have other (negative) consequences. Examples include shorter plays, keeping more tight ends and running backs in, using a fullback or extra tight end / lineman.

[Dolfanalyst]

Again, your analysis completely ignores what actually happens on the field. If a team has a weak line, what do they do? They find other ways to compensate, at least to some extent, for it, even though those actions will have other (negative) consequences. Examples include shorter plays, keeping more tight ends and running backs in, using a fullback or extra tight end / lineman.

And theoretically that's true.  But do you have any evidence that the Dolphins have done that any more than any other team, and if so, how much?

[Dolfanalyst]

Interesting article here that bears on this topic:

http://sports.yahoo.com/news/it-might-be-too-late-for-seahawks-to-fix-their-biggest-flaw-023707190.html

Alas, for the salary cap-strapped Seahawks, it was not to be. They couldn’t get into the Sitton sweepstakes. But the yearning sentiment from that September day is lingering. And it will all season long because the Seahawks have problems that are extremely hard to fix inside an NFL season. And Sunday’s 25-20 loss to a mediocre New Orleans Saints team – while being very poorly officiated – showed how the lack of offensive line investment has manifested itself.

This is going to be a problem in January. Most likely because the Seahawks will likely have to go through the Dallas Cowboys, who are now the biggest offensive bullies in the NFC.

Before that happens, there is plenty to solve in Seattle offensively. Most of it tracing back to the line. Russell Wilson? Beat up. The running game? Lethargic. Big plays downfield? They don’t have the time to develop. And in the middle of it all is an offensive line that lags far behind the rest of the roster in talent and investment.

If you look at Wilson's performance this year, I submit it's a study in how a QB's performance can be diminished by lack of mobility (in this case due to injury) in the face of pressure.

Wilson last year averaged 34 rushing yards per game, and in 2014 53 yards per game.  This year, after injuring his knee in game one, he's averaging just 6 yards per game rushing.

His average QB rating previously in his career was 102.  This year it's just 91.5.

His offensive line has been widely believed to be suspect throughout his career, and although the above article points to its deficiencies this year, the fact remains that he's been sacked roughly half as often as he had been previously in his career.

So if you "reverse engineer" the above, you can determine what might be possible for Ryan Tannehill if he were to continue to show the pass rush awareness and evasion skills he showed in the Buffalo game last week.

Rather than being confined to the pocket and attempting to make plays solely from there, Tannehill could expand his game considerably by being aware of pressure and evading it, buying time for receivers to break free downfield and making plays similar to the clutch one he made with Jarvis Landry late in the game against Buffalo.

This is what Russell Wilson used to do aplenty previously in his career, and I think you can see what becoming more "Tannehill-esque" (again due to injury) has done to his numbers.