Tannehill and the Offensive Line

[Spider-Dan]

Based on YPA, Tannehill's best game that stretch was the one in which he posted the worst passer rating by far; a game in which the OL played incredibly poorly and the offense was completely inept.  Yet in that same stretch, PFF had Tannehill's best game as the one with his second-worst YPA.

[Dolfanalyst]

Based on YPA, Tannehill's best game that stretch was the one in which he posted the worst passer rating by far; a game in which the OL played incredibly poorly and the offense was completely inept.  Yet in that same stretch, PFF had Tannehill's best game as the one with his second-worst YPA.

Tannehill had only one poor game in that stretch in terms of YPA, and that was New York.  The rest ranged from above average (7.9) to exceptional (10.3).  So the game in which he had his second-worst YPA (7.9) wasn't bad at all in terms of YPA.

The problem I have with PFF's ratings is that we have no idea what they're based on.

[Dolfanalyst]

• One of Adam Gase’s missions, beyond transforming this into a winning organization, was making Ryan Tannehill a better quarterback. And in barely more than half a season, Gase already has done it, with Tannehill obviously also deserving a large share of the credit.

Tannehill’s passer rating has risen from 88.7 last season to 91.3, 16th best among all starters and ahead of, among others, Eli Manning (88.4 entering the Giants’ Monday night game), Jameis Winston (87.5), Carson Palmer (86), Cam Newton (86), Joe Flacco (78.3), Jay Cutler (77.1) and Brock Osweiler (74.1).

Tannehill is averaging a career-high 8.0 yards per attempt, better than his career mark of 7.0. He’s completing 65.3 percent of his passes, well ahead of his 62.3 career mark.

• During the past four games, he has a passer rating of more than 97 in three of them – 97.4 against Pittsburgh, 99.4 against Buffalo and 130.6 against San Diego --- and has four touchdown passes and no interceptions. That 130.6 against the Chargers was the second-best of his career.

A huge key: He has been sacked only three times in those four games.

• This also speaks well of Gase: Chicago’s Cutler posted a career-high 92.3 rating in his one year with Gase as his offensive coordinator but has slipped to 77.1 this season without Gase.

The "huge key" of having been sacked only three times in those four games again tends to get people off into the wild goose chase of attributing Tannehill's play to the offensive line, whereas 1) the game-to-game correlation between sacks and QB rating throughout his career is weak (i.e., he's had a fair number of games in which he was sacked far more times and had a high QB rating nonetheless, as well as a fair number of games in which he was sacked minimally and had a low QB rating nonetheless), and 2) the three sacks in four games mentioned could just as easily be attributable to improved awareness and evasion of pressure on his part, i.e., his playing the position better.

BTW, Tannehill's 91.3 QB rating, if it continues, will be the second time in his career he's had an above-average QB rating.  It's not significantly above-average, but it is nonetheless.

Really the best sign from the game yesterday in my opinion is that he put together all of the elements of the position -- aggressive downfield passing (YPA = 10), accuracy (70.8%), ball protection (0 INTs), and passing well under pressure (QB rating under pressure was 155).  It was a complete game.

http://www.miamiherald.com/sports/spt-columns-blogs/barry-jackson/article114720223.html#storylink=cpy

[fyo]

My only comment on Tannehill and the O-line this week is going to be to note that DVOA is going to *kill* Tannehill for the 3rd and long strip sack despite it being 100% on Branden Albert. The sack was helpfully timed by the CBS crew and happened in 2.38 seconds. Getting hit that quickly is always really bad, but on a long drop back, it's a killer. Especially when it happens on the outside and Albert doesn't even get in position to allow Tannehill to step up. Blind-side strip sack. Even a few tenths more and *maybe* Tannehill has a chance to step up into the pocket. Overall, I suspect DVOA will give Tannehill an solid game since he did complete 70+% of his throws, threw 2 touchdowns, 10 yards per attempt, the first down run, and 11 first downs by passing. Still, DVOA counts a fumble as half a turnover, plus the sack will cost as well, so it's going to cost.

I keep meaning to rewatch all of Tannehill's plays and note how he does in different situations. Historically, he's been *elite* when flushed/bootleg and the numbers have felt pretty good this year as well. Certainly in this game he made a lot happen on the plays where he got away from the offensive line. Historically, he's been pretty bad at quick outs from the pocket and I have a feeling that he's still not looking great in that area.

And, not to excuse Rivers, but he played a lot better than his numbers (traditional or DVOA) will show. His receivers were dropping balls all over the place and not making the proper reads.

[Dolfanalyst]

My only comment on Tannehill and the O-line this week is going to be to note that DVOA is going to *kill* Tannehill for the 3rd and long strip sack despite it being 100% on Branden Albert. The sack was helpfully timed by the CBS crew and happened in 2.38 seconds. Getting hit that quickly is always really bad, but on a long drop back, it's a killer. Especially when it happens on the outside and Albert doesn't even get in position to allow Tannehill to step up. Blind-side strip sack. Even a few tenths more and *maybe* Tannehill has a chance to step up into the pocket. Overall, I suspect DVOA will give Tannehill an solid game since he did complete 70+% of his throws, threw 2 touchdowns, 10 yards per attempt, the first down run, and 11 first downs by passing. Still, DVOA counts a fumble as half a turnover, plus the sack will cost as well, so it's going to cost.

And that may be true, but that's problematic with regard to DVOA's validity only if either or both of the following are true:  1) the formula associated with that play is applied unequally among QBs across the league by the folks at Football Outsiders (i.e., when it happens to Tannehill, it's counted against him, but when it happens to Tom Brady for example, he's given a pass), and/or 2) those sorts of plays happen significantly more frequently for some teams than others over the long haul.

The first of those two possibilities is exactly why subjectively determined statistics can be problematic.  It would take superhuman abilities (which none of us have) not to apply subjective appraisal formulas unequally across QBs.  And if that happens to a sufficient degree, then any distinctions among QBs, i.e., the sorts of rankings of them we see all the time, are invalid.

Also, we know that ESPN's QBR involves the appraisal of QB pressure and divides responsibility for the outcomes of plays on the basis of it.  Tannehill had the 10th-best QBR in the league this year at San Diego, so one may prefer that statistic.

[fyo]

There are still more than a few subjective elements to QBR, there is an egregious win-bias, and there's a whole boat-load we don't know anything about, because ESPN haven't shared that information. The bottom line is that, while Tannehill scores well this time around on QBR, I cannot consider it significantly less subjective than, e.g., PFF or Bleacher Reports play-analysis grades (where care *is* taken to apply a consistent standard, although it (like QBR) has subjective elements).

As for quick sacks vs slow sacks, there is absolutely no question that this happens to some quarterbacks more than others. FO themselves did a study and for the season they looked at, no quarterback had more of these quick sacks than Tannehill. Unfortunately, no one publishes weekly sack times, so we're left to wonder at magnitude of the effect.

[Dolfanalyst]

There are still more than a few subjective elements to QBR, there is an egregious win-bias, and there's a whole boat-load we don't know anything about, because ESPN haven't shared that information. The bottom line is that, while Tannehill scores well this time around on QBR, I cannot consider it significantly less subjective than, e.g., PFF or Bleacher Reports play-analysis grades (where care *is* taken to apply a consistent standard, although it (like QBR) has subjective elements).

As for quick sacks vs slow sacks, there is absolutely no question that this happens to some quarterbacks more than others. FO themselves did a study and for the season they looked at, no quarterback had more of these quick sacks than Tannehill. Unfortunately, no one publishes weekly sack times, so we're left to wonder at magnitude of the effect.

I don't mean to sound combative, but even the variable of quick versus slow sacks likely has variance attributable to things other than the offensive line.  A couple come to mind immediately:  1) how often teams are in obvious passing situations, which lets opposing defenses tee off on the offensive line and the passer, and 2) how well QBs recognize blitzers pre-snap and evade them post-snap.

Personally I come back, once again, to the idea that there is very little variation across the league in how often teams produce a "clean pocket" for the QB (i.e., ESPN's study I mentioned above).  When that's the case, it becomes difficult for me to attribute differences in QB play to offensive lines.  There just isn't the variance in offensive line play needed to account for the variance in QB play.

[fyo]

I don't mean to sound combative, but even the variable of quick versus slow sacks likely has variance attributable to things other than the offensive line.  A couple come to mind immediately:  1) how often teams are in obvious passing situations, which lets opposing defenses tee off on the offensive line and the passer, and 2) how well QBs recognize blitzers pre-snap and evade them post-snap.

Personally I come back, once again, to the idea that there is very little variation across the league in how often teams produce a "clean pocket" for the QB (i.e., ESPN's study I mentioned above).  When that's the case, it becomes difficult for me to attribute differences in QB play to offensive lines.  There just isn't the variance in offensive line play needed to account for the variance in QB play.

There is next to nothing a quarterback can do to evade a sack that happens after 2.38 seconds on a long drop with the blind side tackle getting beat 1-on-1 by his guy. This wasn't because of a blitz, this was just Albert getting beat badly, resulting in a strip-sack.

The ESPN study you linked doesn't come to the same conclusions that you do, doesn't state what its objective criteria were, and doesn't consider what kind of help the offensive line had in keeping the pocket clean.

One of the problems we continuously face is that there are no good, objective measures of line play. Sack speed would be one possible objective measure. For run-blocking you could look at avg running back yards before first contact. ESPN and PFF both chart the latter, but neither make the numbers available.

[Dolfanalyst]

There is next to nothing a quarterback can do to evade a sack that happens after 2.38 seconds on a long drop with the blind side tackle getting beat 1-on-1 by his guy. This wasn't because of a blitz, this was just Albert getting beat badly, resulting in a strip-sack.

Right, and so on that particular play, the quick sack was attributable to the offensive line.  That doesn't mean all or even the majority of quick sacks are attributable to offensive lines.  At best it's a variable with multiple sources of variance, the degree of each is unknown.

The ESPN study you linked doesn't come to the same conclusions that you do, doesn't state what its objective criteria were, and doesn't consider what kind of help the offensive line had in keeping the pocket clean.

I don't need ESPN to tell me that less than 10% variance between the best and the worst team in the league in terms of frequency of clean pockets is comparatively miniscule.

But yes, as with the point I'm making with regard to "quick sacks," there are likely multiple sources of variance with regard to the frequency of clean pockets provided.

One of the problems we continuously face is that there are no good, objective measures of line play. Sack speed would be one possible objective measure. For run-blocking you could look at avg running back yards before first contact. ESPN and PFF both chart the latter, but neither make the numbers available.

Yes that's an issue.  And it's also an issue that quarterback play is weakly related to the measures we do have, such as sacks and frequency of QB pressure.

[fyo]

Right, and so on that particular play, the quick sack was attributable to the offensive line.  That doesn't mean all or even the majority of quick sacks are attributable to offensive lines.  At best it's a variable with multiple sources of variance, the degree of each is unknown.

Everything in football is a variable with multiple sources of variance, each of unknown magnitude. The same argument could be made about a quarterback throwing an incomplete. Was that a bad throw? Bad route? Something else?

I don't need ESPN to tell me that less than 10% variance between the best and the worst team in the league in terms of frequency of clean pockets is comparatively miniscule.

With no methodology, no "time" resolution, and no differentiation between a 6 and 8 man rush, or between 5 offensive linemen and max protect, the vague ESPN numbers are fairly useless.

[Phishfan]

Everything in football is a variable with multiple sources of variance, each of unknown magnitude.

Ding Ding Ding

[Dolfanalyst]

Everything in football is a variable with multiple sources of variance, each of unknown magnitude. The same argument could be made about a quarterback throwing an incomplete. Was that a bad throw? Bad route? Something else?

For small sample sizes that's true, but there must be some way we arrive at the conclusion that a player is "good," and that's done intuitively via larger sample sizes.  At that point it becomes possible to tease apart the various influences on his play.

Nobody is still wondering whether Dan Marino was good, for example, and concluding that "there were indeterminable sources of variation in his play, each of unknown magnitude, and so we really have no idea how good he was."

[Dolfanalyst]

Should be interesting this week to see how Tannehill performs with Albert and possibly Pouncey out.  I suspect that his success will hinge on how well he continues to move in response to pressure.

[CF DolFan]

Should be interesting this week to see how Tannehill performs with Albert and possibly Pouncey out.  I suspect that his success will hinge on how well he continues to move in response to pressure.

It depends on how much time he has to move. Our o-line doesn't just struggle. At times it doesn't even slow down the opposition. Fortunately it isn't Thomas and Turner as the back-ups.

[Dolfanalyst]

It depends on how much time he has to move. Our o-line doesn't just struggle. At times it doesn't even slow down the opposition. Fortunately it isn't Thomas and Turner as the back-ups.

Right, but he also needs to understand that he'll be working without one and possibly two starting offensive linemen, and so moving in response to pressure should be more at the forefront of his mind than usual.

This is the same thing that should happen for him when he's in an obvious passing situation in game and more likely to face increased pressure that possibly overwhelms the offensive line.

There are times when a QB can devote more of his attention downfield, and other times when he should devote more of it to what's going on with the pass rush.  This of course is called "situational awareness," and is part of playing the quarterback position.  Anybody who hopes to play the position well needs it.