2014 Pressure Plays: Quarterbacks
by Sterling Xie
Quarterback evaluation tends to receive more scrutiny than the Zapruder film. And while the position's magic bullet formula may never reveal itself, one universally accepted maxim is that a quarterback must be able to handle pressure. In conjunction with ESPN Stats & Info, the Football Outsiders game charting project annually tracks pass pressure results. Starting with offensive pressure numbers, we'll take a look at these charting results, which will be available in Football Outsiders Almanac 2015 next month.
Penned by Scott Kacsmar, last year's rendition of this article found a lot of inconsistencies in how quarterbacks perform under pressure from year to year. Part of this stems from sample size, since we're only dealing with a quarter of each passer's plays, and the general havoc of a typical pressure play lends itself to wider variation over time. Even the amount of pressure a quarterback sees doesn't tend to affect how he performs with pass rushers in his face. For 2014, the correlation between pressure DVOA and pressure rate was -0.16, suggesting only a tiny negative relationship between the two stats.
A couple commenters last year wondered about pressure in conjunction with blitz rates. At first glance, it appears that blitzing is generally more effective than normal plays. In 2014, defenses generated pressure on 33.9 percent of all blitzes (defined here as any play with five or more pass rushers), while schemes with four or fewer rushers generated pressure only 20.4 percent of the time. The DVOA gap between blitzes and non-blitzes, though, was much closer. This is mostly because defenses that can generate pressure with a four-man rush benefit from the extra defenders left to play pass coverage. The following table shows how offenses fared vs. blitzes and non-blitzes in 2014:
| 2014 Offensive DVOA vs. Pressures and Blitzes | ||||
| Play | Overall DVOA | Pressure Rate | DVOA, Pressure | DVOA, No Pressure |
| Blitz | 10.1% | 33.9% | -67.8% | 47.8% |
| Non-Blitz | 17.5% | 20.4% | -84.4% | 42.5% |
Comments
21 comments, Last at 29 Jun 2015, 1:29pm
#1 by jtr // Jun 25, 2015 - 3:04pm
Sterling, I like your writing, looking forward to reading more from you on the site.
I'd be curious to see the air yards/YAC breakdown on pressured QBs. Roethlisburger is still hard to bring down, but he doesn't run around for 10 seconds breaking tackles as often under Haley as he did under Arians. I have a feeling that his excellent DVOA under pressure this season is largely related to the fact that he has two excellent YAC artists in Bell and Brown who can each turn a short pass into a big gain. That would particularly make sense for the success on blitzes, where a checkdown receiver often finds himself with a lot of space to work with.
#2 by speedegg // Jun 25, 2015 - 3:26pm
Interesting observation about Carson Palmer. I think Doug Farrar (or Greg Cosell?) mentioned that Palmer has "read it out" on his reads for passing plays. So, he was almost robotic and not improvisational like Roethlisberger or quick reacting like Brees. So was last year a fluke or if Arians changed the way he teaches his system/reads or if Palmer became more comfortable with the offense
#5 by MilkmanDanimal // Jun 25, 2015 - 5:26pm
I saw the article's headline and immediately put my head down on the desk. "Josh McCown will be last," I said, "but the long nightmare from last season is over. I won't have to relive it."
THANKS FOR THE THAT GIF GUYS, I REALLY NEEDED TO REMEMBER THAT.
2014 never happened.
#6 by theslothook // Jun 25, 2015 - 5:59pm
Funny thing was, there wasn't a play the rest of the season that was quite the same. In a separate thread, I rehashed my personal favorites. I still maintain even after seeing that play a dozen times over...aaron brooks' worse was worse.
#9 by theslothook // Jun 25, 2015 - 7:12pm
Technically, its determined by espn before the charters get to it. That said, as a charter, if I see something that was a pressure but wasn't marked, I can mark it as missed.
Determining pressure i suppose is pretty subjecting. I tend to mark cases when a pass rusher has beaten his man.
#11 by Topas // Jun 26, 2015 - 4:07am
I dont understand that part:
> Unsurprisingly, our initial two observations hold up: Offenses fare better against the blitz regardless of whether or not the pressure gets there.
The DVOA against Blitzes is 10.1. Against non-Blitzes it is 17.5. So offenses do fare better if the defense does not blitz. But you state the opposite. Am I missing something?
#12 by jtr // Jun 26, 2015 - 7:37am
He's referring to the fact that Pressure DVOA and Non-Pressure DVOA are both higher for blitzes than for non-blitzes. Since the pressure reaches home more often on blitzes, overall blitz DVOA contains more of the former and less of the latter, so the overall DVOA is lower for blitzes. It does kind of make sense that once you control for whether or not the pressure gets there, defenses do better with more guys in coverage. But that fact is more than negated by the obvious fact that pressure arrives more often on blitzes than on seven-man-drops.
#15 by dank067 // Jun 27, 2015 - 12:12pm
Another fun one from this table: Russell Wilson is 5th in DVOA vs pressure and 3rd in DVOA without pressure, but only 14th in total passing DVOA. (These numbers include scrambles, so not identical data sets, but I imagine they're still close.)
This is kind of an interesting case when trying to project Wilson moving forward. On one hand, if Wilson/the OL/the offensive scheme can combine to reduce pressure rate, Wilson could be in line for a nice boost in efficiency. But unless he can keep up his relative success vs. pressure, he also appears to be at risk for a dropoff as well.
#16 by MC2 // Jun 27, 2015 - 10:59pm
These numbers include scrambles, so not identical data sets, but I imagine they're still close.
With most QBs, that would probably be a safe assumption, but considering how great Wilson's rushing numbers were last year (269 rushing DYAR, more than the next 2 guys combined), I'm guessing scrambles did play a significant role in that discrepancy.
#17 by dank067 // Jun 28, 2015 - 1:22am
Maybe a little bit, but probably not enough to move him up more than a couple spots. If you rough it out and add Wilson's entire rushing DYAR to his passing DYAR, looks like he would only move from 13th (in QB DYAR) to 11th. And even then, out of the 97 runs FO has him listed for, not all of those were scrambles.
#20 by Perfundle // Jun 29, 2015 - 12:02am
That's DYAR, not DVOA. With his pass attempts so low and his scramble attempts so high adding in the scrambles will really boost his numbers. Also, his non-scramble runs are considerably less successful than his scrambles, which pushes his pass+scramble DVOA up even higher. If you calculate each QB's DVOA weighted by pressure (e.g. Wilson's is 75.3%*(1 - 39.1%) - 44.5%*39.1% = 28.5%), Wilson is 8th.
#21 by dank067 // Jun 29, 2015 - 1:29pm
Seattle actually finished 10th in total passing offense DVOA last season at 21.0%, so it looks like including scrambling moves them up about two spots and about 7.5% DVOA in that regard as well. I now wonder how passing offense DVOA and QB passing DVOA differ, because as mentioned earlier Wilson was 14th in passing DVOA alone at 5.5%. (Maybe passing offense DVOA already includes scrambles?)
Also, I think combining DYAR is still reflective of Wilson's scrambling efficiency. He produced 296 rushing DYAR on 97 runs, compared to 503 passing DYAR on 495 plays, or about 3/5 as much value on 1/5 as many plays. (Unless the baselines for passing and rushing DYAR are so different this wouldn't make sense.)
#18 by jtr // Jun 28, 2015 - 1:21pm
Roethlisburger would be a reason for optimism with Wilson. Big Ben has consistently been near the top in pressure DVOA using a similar approach of scrambling around behind the line of scrimmage to buy time for a receiver to get open. When it works, it really really works since there's no way coverage can hang on for 6+ seconds, especially when you're trying to cover your man/zone while simultaneously helping contain the QB.
#19 by Parmenides // Jun 28, 2015 - 4:55pm
Is there a charting that separates quick pressure and long pressure? That would seem to separate out some qb effects and line effects. Its one thing if the pressure results after 3 seconds and another if its immediately after the snap. Another idea would be to separate pressure as middle or edge.