What a difference a line makes.

[Spider-Dan]

How exactly do we know when to make one of the following conclusions:

1) The QB is playing well (or poorly) because of those around him.
2) The QB is playing well (or poorly) because of his own ability (or the lack thereof).
3) Those around the QB are playing well (or poorly) because of the QB.

There are answers to these questions.

MIA's offensive line grades out poorly in almost every pass protection metric.  You can complain all you like about a QB not "feeling pressure," but an OLineman's job is not to get bowled over in the first 3 seconds; if he's not making his block, he's not doing his job.

Similarly, if a QB is getting drops and/or interceptions because of poorly placed throws, it's fair to blame that on him and not his receivers.  But when he is putting the football where it needs to be and his receiver is just dropping it, or when (according to the coaching staff that calls the plays) the receiver runs the wrong route, this is not the fault of the QB.

Fundamentally, your position is contradictory; you swiftly and repeatedly come to conclusive evaluations of the QB, yet for virtually any other position you insist evaluation is impossible because we can't know how much we should attribute unacceptable performance to that individual person's play.  If we can't grade an OLine as bad because we don't know how much the QB is contributing to their play, how can we possibly grade the QB in the reverse?

[Dolfanalyst]

There are answers to these questions.

MIA's offensive line grades out poorly in almost every pass protection metric.  You can complain all you like about a QB not "feeling pressure," but an OLineman's job is not to get bowled over in the first 3 seconds; if he's not making his block, he's not doing his job.

Similarly, if a QB is getting drops and/or interceptions because of poorly placed throws, it's fair to blame that on him and not his receivers.  But when he is putting the football where it needs to be and his receiver is just dropping it, or when (according to the coaching staff that calls the plays) the receiver runs the wrong route, this is not the fault of the QB.

Fundamentally, your position is contradictory; you swiftly and repeatedly come to conclusive evaluations of the QB, yet for virtually any other position you insist evaluation is impossible because we can't know how much we should attribute unacceptable performance to that individual person's play.  If we can't grade an OLine as bad because we don't know how much the QB is contributing to their play, how can we possibly grade the QB in the reverse?

Good question (the bolded portion).  Because there are statistics available 1) that distinguish good from bad QBs, which gives them validity as measures of QB play, and 2) that don't correlate strongly with any known measures of offensive line play, other offensive play (running game), or defensive play.

[Spider-Dan]

[re: how can we rate QBs but not OLines?]

Because there are statistics available 1) that distinguish good from bad QBs, which gives them validity as measures of QB play

It appears that you just said that we can rate good and bad QBs because there are statistics that distinguish good from bad QBs, which gives said statistics validity.
That is circular reasoning.

If we cannot evaluate and grade OLines because their (poor) performance might be influenced by other factors, it is contradictory to say that we can isolate and evaluate QB play, when QBs objectively have many more factors contributing to the quality of their play than OLinemen do.  Again, when the OL is being bowled over instantly or allows a defender to run right past them, the QB play is irrelevant to that player's performance.

[Dolfanalyst]

[re: how can we rate QBs but not OLines?]
It appears that you just said that we can rate good and bad QBs because there are statistics that distinguish good from bad QBs, which gives said statistics validity.
That is circular reasoning.

The statistics themselves don't determine the good and bad QBs.  That's based on consensus, rather.  This is no different from saying a thermometer is a valid measurement of temperature because it shows a higher reading when something is warm to the touch than it does when something is cool to the touch.

If we cannot evaluate and grade OLines because their (poor) performance might be influenced by other factors, it is contradictory to say that we can isolate and evaluate QB play, when QBs objectively have many more factors contributing to the quality of their play than OLinemen do.  Again, when the OL is being bowled over instantly or allows a defender to run right past them, the QB play is irrelevant to that player's performance.

Again we can isolate QB play because the statistics that measure QB play correlate weakly with all other known measures of NFL play.

So yes, we can attribute pressure and/or a sack caused by the situation you mentioned to the offensive line, for example, but QB statistics don't correlate strongly with pressure and sacks.

[fyo]

Again we can isolate QB play because the statistics that measure QB play correlate weakly with all other known measures of NFL play.

Unless you can provide a measure of line play, you cannot make the above claim. What you are measuring by your QB statistic is then at best QB+Line.

In fact, there are many factors that contribute to the QB metrics you like to use and there just isn't enough data available, at least that I've seen, to really get a good separation of contributions.

Take Yards After Catch, as an example. The Dolphins have had sensationally poor YAC production from their receiving corps since even before Tannehill got here. Now, is YAC more a product of the receiver or of the quarterback? Logically, you could argue it either way and even looking (at least superficially) at the statistics it is muddled (one problem is that league average YAC varies significantly dependent on the receiver route, but no publicly available source of YAC lists ANY accompanying data). If quarterback were the main component in YAC (argument being ball placement and timing), then new receivers shouldn't change much. However, when Jarvis Landry joined the Dolphins, he immediately started racking up serious YAC. The majority of the top 10 YAC producers over the past decade have been running backs and tight ends (8 out of the 10 top spots for 2016 so far), however Miami has not had a RB or TE crack the top 40 during Tannehill's tenure. The result being that even with Landry's YAC production, Miami is still (for whatever reason) getting a total YAC contribution that is significantly below league average.

When looking at sacks, and I've stated this before, there's a significant difference between sacks that happen within the first 2.4 seconds and sacks that take a long time. Those quick sacks are pretty exclusively on the offensive line (for a given play call). The only data we have on this, which covers less than one full season, point to the Dolphins having a ridiculous number of these quick sacks compared to other teams and a relatively low number of slow sacks (second fewest slow sacks, third highest short sacks, among players who played at least 50% of games).

There are other aspects that have to average out over a full season for the statistics to actually be significant. However, that assumption is not valid for the low numbers and high variances that underpin many (if not most) of the metrics we use. The number of passes dropped, for instance, does not average out over a season (and appears to be very dependent on receiver as opposed to quarterback). Even more so for dropped touchdown passes (like the two this Sunday), or indeed any other sub-catagory of dropped passes. Receivers running wrong routes is something one would expect to occur more often, on average, with new coaching staff and new receivers, and would contribute both to interceptions and lower production in other passing metrics.


So in Tannehill, specifically, we have:

1. A quarterback who, on average, has produced an average to somewhat below average overall performance.
2. A quarterback who has elite production when out of the pocket, both on designed rollouts and on scrambles.
3. A quarterback who has an offensive line that has consistently been ranked among the worst in the league, e.g. by PFF (and supported by the objective, if very limited short sack stats).
4. A quarterback who has seen significantly below average YAC production from those catching his passes, even with Jarvis Landry.
5. A quarterback who has had a significantly higher number of changes in coaching staff than average.

How do you evaluate such a quarterback?

He might be just an average quarterback -- and I really do believe the combined evidence makes it very unlikely he is below that -- or he could be significantly better. How do we tell where in this range his actual ability falls?

[Dolfanalyst]

Unless you can provide a measure of line play, you cannot make the above claim. What you are measuring by your QB statistic is then at best QB+Line.

In fact, there are many factors that contribute to the QB metrics you like to use and there just isn't enough data available, at least that I've seen, to really get a good separation of contributions.

Take Yards After Catch, as an example. The Dolphins have had sensationally poor YAC production from their receiving corps since even before Tannehill got here. Now, is YAC more a product of the receiver or of the quarterback? Logically, you could argue it either way and even looking (at least superficially) at the statistics it is muddled (one problem is that league average YAC varies significantly dependent on the receiver route, but no publicly available source of YAC lists ANY accompanying data). If quarterback were the main component in YAC (argument being ball placement and timing), then new receivers shouldn't change much. However, when Jarvis Landry joined the Dolphins, he immediately started racking up serious YAC. The majority of the top 10 YAC producers over the past decade have been running backs and tight ends (8 out of the 10 top spots for 2016 so far), however Miami has not had a RB or TE crack the top 40 during Tannehill's tenure. The result being that even with Landry's YAC production, Miami is still (for whatever reason) getting a total YAC contribution that is significantly below league average.

When looking at sacks, and I've stated this before, there's a significant difference between sacks that happen within the first 2.4 seconds and sacks that take a long time. Those quick sacks are pretty exclusively on the offensive line (for a given play call). The only data we have on this, which covers less than one full season, point to the Dolphins having a ridiculous number of these quick sacks compared to other teams and a relatively low number of slow sacks (second fewest slow sacks, third highest short sacks, among players who played at least 50% of games).

There are other aspects that have to average out over a full season for the statistics to actually be significant. However, that assumption is not valid for the low numbers and high variances that underpin many (if not most) of the metrics we use. The number of passes dropped, for instance, does not average out over a season (and appears to be very dependent on receiver as opposed to quarterback). Even more so for dropped touchdown passes (like the two this Sunday), or indeed any other sub-catagory of dropped passes. Receivers running wrong routes is something one would expect to occur more often, on average, with new coaching staff and new receivers, and would contribute both to interceptions and lower production in other passing metrics.


So in Tannehill, specifically, we have:

1. A quarterback who, on average, has produced an average to somewhat below average overall performance.
2. A quarterback who has elite production when out of the pocket, both on designed rollouts and on scrambles.
3. A quarterback who has an offensive line that has consistently been ranked among the worst in the league, e.g. by PFF (and supported by the objective, if very limited short sack stats).
4. A quarterback who has seen significantly below average YAC production from those catching his passes, even with Jarvis Landry.
5. A quarterback who has had a significantly higher number of changes in coaching staff than average.

How do you evaluate such a quarterback?

He might be just an average quarterback -- and I really do believe the combined evidence makes it very unlikely he is below that -- or he could be significantly better. How do we tell where in this range his actual ability falls?

It isn't easy.

But let me ask you this:  if someone held a gun to your head (heaven forbid) and said, "put everything you own on what you believe to be the most accurate appraisal of Ryan Tannehill's individual ability as an NFL QB, or I'll pull the trigger," and you knew some omniscient being would then provide the correct appraisal, thus determining what you would win or lose, what would you do?

I can tell you what I'd do, personally, for what it's worth.  I'd go look at the statistics that are best associated with consensus appraisals of QBs' ability, and that correlate least with other measures of NFL play, and use those in my answer.

Now, if I wanted to be extra thorough, I'd also take a look at all the sorts of potential moderating variables you mentioned, and determine the percentage of truly great QBs in NFL history who suffered through four-plus years of such moderation of their individual ability, only to blossom to their significantly higher level of individual play later on.

And that's really the missing piece here, isn't it.  We have this conceptual model of Tannehill's possible trajectory as a player, where he suffers for years in this way, victimized by his surroundings, only to have his surroundings improve and his tremendous individual ability eventually shine through.

But when has that happened for anyone?  Who is the model NFL quarterback(s) that makes such a player trajectory more than just a pipe dream for hopeful (and of course biased) fans of the Miami Dolphins (myself included)?

Or is Tannehill the first NFL quarterback ever for whom this is happening?

The trigger is about to be pulled.  What are you putting your money on?

[fyo]

It isn't easy.

No, it isn't.

But let me ask you this:  if someone held a gun to your head (heaven forbid) and said, "put everything you own on what you believe to be the most accurate appraisal of Ryan Tannehill's individual ability as an NFL QB, or I'll pull the trigger," and you knew some omniscient being would then provide the correct appraisal, thus determining what you would win or lose, what would you do?

Probably shit myself

I can tell you what I'd do, personally, for what it's worth.  I'd go look at the statistics that are best associated with consensus appraisals of QBs' ability, and that correlate least with other measures of NFL play, and use those in my answer.

Now, if I wanted to be extra thorough, I'd also take a look at all the sorts of potential moderating variables you mentioned, and determine the percentage of truly great QBs in NFL history who suffered through four-plus years of such moderation of their individual ability, only to blossom to their significantly higher level of individual play later on.

And that's really the missing piece here, isn't it.  We have this conceptual model of Tannehill's possible trajectory as a player, where he suffers for years in this way, victimized by his surroundings, only to have his surroundings improve and his tremendous individual ability eventually shine through.

But when has that happened for anyone?  Who is the model NFL quarterback(s) that makes such a player trajectory more than just a pipe dream for hopeful (and of course biased) fans of the Miami Dolphins (myself included)?

Or is Tannehill the first NFL quarterback ever for whom this is happening?

The trigger is about to be pulled.  What are you putting your money on?

I'd have to go with my combined impression, which includes all sorts of stats and analysis, plus tons of tape (although not nearly as much as a coach would have access to) and I'd call Tannehill an average starting quarterback with significant upside and little downside.

As for the QB career trajectories you reference, the problem here is really one of sample size (again). Assuming we need at least, say, 5 years to gauge a career, how many quarterbacks have actually been a starter for that long in the past 20 years? (Rule changes make me reluctant to go too far back.) With 32 teams that's a theoretical maximum of 128 careers over that span. The actual number is almost certainly significantly below 100. And now we have to look at career trajectories among those and find the set that most closely matches that of Tannehill and then project his future performance from their future performance?

Of those...

How many quarterbacks do you think have had four offensive coordinators in their first 5 years?
How many quarterbacks do you think have been sacked for even 150+ times (Tannehill 184) in their first four years?
How many quarterbacks do you think have thrown for 14000+ yards (Tannehill 15460) in the first four years?
How many quarterbacks do you think have an td/interception ration similar to Tannehill's (3:2)?
How many quarterbacks have a similar progression of quarterback rating? (Which appears the main stat people use when doing these projections)

Career projection with such a small population without taking individual situation into account is almost certainly going to be futile and completely without statistical significance.

For fun I tried to look at football outsider's "similarity scores" as provided on the player pages, for non-members, and it lists two players: a 2014 Peyton Manning for single season top similarity score and Andy Dalton for career.

[Pappy13]

How exactly do we know when to make one of the following conclusions:

1) The QB is playing well (or poorly) because of those around him.
2) The QB is playing well (or poorly) because of his own ability (or the lack thereof).
3) Those around the QB are playing well (or poorly) because of the QB.

Well it's not easy which is what people have to wrap their head around first off. But you start by looking at the tape and letting your eyes tell you what you see and not only evaluating the QB, but the play around him as well. Then you look at the stats to see trends and see how that meshes with what you see on tape and what conclusions you can make from that. Then you look at practice sessions and see if players make the same mistakes over and over or do they learn from mistakes and make adjustments and use what they are being taught. Then you talk with some others who have done the same things you have just done and see if their opinion is different from yours and if so, see if can determine why and come to a consensus.

What you DON'T do is consider things like salary or draft position which is a trap that many fall into when doing your evaluation. Now once you have done your evaluation then you can do things like consider if their salary is commensurate with their play and try to adjust that accordingly, but that's not part of the evaluation of the player.

What you also DON'T do is consider wins and losses which is another trap. Certainly QB play influences wins and losses, but there's so much more that goes into it that it's not fair to evaluate a player based on team performance.

In short you do what guys like Gase have already done. Now you can listen to what Gase has to say and you can poo-poo it if you want and come up with your own evaluation, but I think it's wiser to listen to him and listen to what he's saying and at least take his opinion into consideration. Gase has defended Tannehill and blasted some of the other players even cutting players. I think that should tell you a lot. You can ignore him if you want, but frankly I think that's unwise.

[Pappy13]

There's one other thing that I'd like to point out and I think what Gase has done with Tannehill is appropriate which is you MAKE him your general on the field. Unlike Philbin, Gase has given Tannehill all the tools and all the support he needs to let everyone on the field know that HE's the general. What he says goes, without question. If he's wrong on the field then he answer's to the coach, but the players aren't allowed to question Tannehill. The QB position DEMANDS that. He MUST have the respect of the players and they must believe in him. If they don't then his chances of leading them are severely diminished. I think that's been a problem in Miami and I think Gase is changing that attitude. I'm not saying that's what's been missing that everything will just fall into place now, Tannehill still has to go out and perform, but it's one thing that has been missing in Miami and has allowed this team to second guess their QB which is a really bad team dynamic. One thing that all good QB's have is the respect of their teammates and I think that has been lacking in Miami partly due to the coaching staff. Gase is changing that and it's for the better.

[Spider-Dan]

Turnabout is fair play:  if someone held a gun to your head (heaven forbid) and said, "put everything you own on what you believe to be the most accurate appraisal of the performance of the Dolphins' offensive line over the last 4 seasons, or I'll pull the trigger," and you knew some omniscient being would then provide the correct appraisal, thus determining what you would win or lose, what would you do?

Before we can even have a discussion about whether the line matters (which is supposed to be the topic of this thread), we have to first agree that evaluating the performance of the line is a thing that is possible.  After we have evaluated the line, if you want to say that you don't think the OL matters when evaluating how a QB has played, I'm happy to let that statement stand on its own merit.

[Spider-Dan]

The statistics themselves don't determine the good and bad QBs.  That's based on consensus, rather.

So why can't we determine good and bad OLs based on "consensus"?

This is no different from saying a thermometer is a valid measurement of temperature because it shows a higher reading when something is warm to the touch than it does when something is cool to the touch.

Thermometers simply provide factual measurements of temperature.  Warm/cool, like good/bad, are subjective assessments; you are effectively claiming that it's impossible to know whether a line is bad, but it's readily apparent whether a QB is bad.

To complete your analogy, when in comes to OLs, you say, "How can any of us truly know what warm really is?", but when it comes to QBs, you say, "This guy is definitely extremely cold."

[Dolfanalyst]

No, it isn't.

Probably shit myself


I'd have to go with my combined impression, which includes all sorts of stats and analysis, plus tons of tape (although not nearly as much as a coach would have access to) and I'd call Tannehill an average starting quarterback with significant upside and little downside.

And I would actually agree with that, believe it or not.

As for the QB career trajectories you reference, the problem here is really one of sample size (again). Assuming we need at least, say, 5 years to gauge a career, how many quarterbacks have actually been a starter for that long in the past 20 years? (Rule changes make me reluctant to go too far back.) With 32 teams that's a theoretical maximum of 128 careers over that span. The actual number is almost certainly significantly below 100. And now we have to look at career trajectories among those and find the set that most closely matches that of Tannehill and then project his future performance from their future performance?

Of those...

How many quarterbacks do you think have had four offensive coordinators in their first 5 years?
How many quarterbacks do you think have been sacked for even 150+ times (Tannehill 184) in their first four years?

The problem with that statistic, per se, is that 1) sacks don't correlate strongly with QBs' performance in the NFL, and 2) Tannehill's performance, specifically, doesn't correlate strongly with sacks on a game-by-game basis.  He's had a good number of better games in which he's been sacked comparatively more, and a good number of poorer games in which he's been sacked comparatively less.

I hear this thing about Tannehill's "sacks" mentioned all the time, however, even by the media, as though those correlations are strong, whereas in fact they aren't.

How many quarterbacks do you think have thrown for 14000+ yards (Tannehill 15460) in the first four years?
How many quarterbacks do you think have an td/interception ration similar to Tannehill's (3:2)?
How many quarterbacks have a similar progression of quarterback rating? (Which appears the main stat people use when doing these projections)

Career projection with such a small population without taking individual situation into account is almost certainly going to be futile and completely without statistical significance.

For fun I tried to look at football outsider's "similarity scores" as provided on the player pages, for non-members, and it lists two players: a 2014 Peyton Manning for single season top similarity score and Andy Dalton for career.

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.

[Dolfanalyst]

So why can't we determine good and bad OLs based on "consensus"?
Thermometers simply provide factual measurements of temperature.  Warm/cool, like good/bad, are subjective assessments; you are effectively claiming that it's impossible to know whether a line is bad, but it's readily apparent whether a QB is bad.

To complete your analogy, when in comes to OLs, you say, "How can any of us truly know what warm really is?", but when it comes to QBs, you say, "This guy is definitely extremely cold."

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.  I suspect there would be so much variation in people's impressions of offensive line play from team to team that no consensus would be possible, whereas with QBs that wouldn't happen.

To put it back in the original language, too many people would call the same offensive lines both "cold" and "hot" (varying from person to person), that no consensus would be possible.  With QBs I don't suspect that would happen, however.

[Dolfanalyst]

Turnabout is fair play:  if someone held a gun to your head (heaven forbid) and said, "put everything you own on what you believe to be the most accurate appraisal of the performance of the Dolphins' offensive line over the last 4 seasons, or I'll pull the trigger," and you knew some omniscient being would then provide the correct appraisal, thus determining what you would win or lose, what would you do?

I would say it's average to non-significantly below average.  First, I don't think there is as much variation in offensive lines across the league as some appear to believe, and second, I don't think the Dolphins' line has been as poor in comparison to the average NFL offensive line as some have portrayed it to be.

Remember that offensive lines comprise the single biggest unit of players in football.  As you increase the number of players in a unit, the parity the league is driven by increases, and there is a greater chance for some players in the unit to compensate for the play of others.  If that's the case, then there will be comparatively little variation across lines in the NFL.

And if there is comparatively little variation across lines, it's impossible for any line to be a whole lot worse than any other.  All offensive lines would hover around a relatively narrow range of play.

Before we can even have a discussion about whether the line matters (which is supposed to be the topic of this thread), we have to first agree that evaluating the performance of the line is a thing that is possible.  After we have evaluated the line, if you want to say that you don't think the OL matters when evaluating how a QB has played, I'm happy to let that statement stand on its own merit.

It matters, but the question is, how much, and an additional question is, how much does the QB matter in evaluating the offensive line's play?

[Phishfan]

and there is a greater chance for some players in the unit to compensate for the play of others. 

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.