2013 Pressure Plays: Quarterbacks
by Scott Kacsmar
Under pressure, pushing down on you, no quarterback asks for it. The Football Outsiders game charting project, in conjunction with ESPN Stats & Information, tracks pass pressure, along with many other things. We've published the results in Football Outsiders Almanac 2014 (can be purchased here). In the first of a two-part series on the impact of pressure in 2013, we look at how the quarterbacks fared.
One of the silliest statements you will hear on TV is when an analyst says "[Quarterback X] is a different guy when he's pressured." They usually pick one of the best players in the league to make it sound like they've cracked his code, but this statement can be applied to every single quarterback in the NFL. We have four years of this data and every quarterback with a qualified number of attempts has worse statistics when pressure is present.
The value-added analysis is finding out how each quarterback handled the pressure. The strange part is this fluctuates from year to year. In 2012, Andrew Luck (-80.2%) and Ben Roethlisberger (-80.7%) had the smallest declines in DVOA with pressure compared to without pressure. Makes a lot of sense, right? They are both big guys, hard to bring down and mobile enough to get out of trouble. This past season, Luck (-102.5%) fell to 15th and Roethlisberger (-111.3%) to 19th. This is a stat where most would expect Aaron Rodgers to shine as well, because he can throw with the best of them, he can do it on the run, and he can reset his feet and fire accurately. Well, after ranking 10th in "pressure drop" in 2012, Rodgers had the third-worst DVOA decline in the league when under pressure last year.
I first noticed the lack of year-to-year consistency when I started collecting general stats on "Under Pressure" splits from STATS LLC over eight years ago. Not helping matters is the fact that pressure can be a very subjective "eye of the beholder" stat to track. We have plays where we disagree with ESPN on the presence of pressure, and we suggest revisions all the time. If there's a disagreement about the amount of pressure on a 50-yard touchdown pass, that's going to skew the season stats.
So maybe it's due to the difficulties of game charting, or maybe pressure just creates a more chaotic game with bigger gains and bigger losses. Receivers break their routes, defenders lose their assignments, and you can have a wide-open receiver for a huge gain. You could also suffer a 12-yard loss on a sack. Pressure really changes the game.
The following table shows all 40 quarterbacks with at least 200 regular-season pass plays. Scrambles and Defensive Pass Interference are included, but aborted snaps are excluded. The quarterbacks are sorted by ascending frequency of pressure, and the "Difference" section is ranked from smallest to largest in terms of the drop in DVOA when under pressure.
A couple of housekeeping notes: First, sacks marked as "coverage sack" and scrambles marked as either "coverage scramble" or "hole opens up" do not count as pressure plays. Second, DVOA here is not actually passing DVOA; it is team offense DVOA with this player either passing or scrambling. The reason for that is that scrambles are counted as runs rather than passes in DVOA. That's something we plan on eventually changing, but we haven't had the opportunity yet (and of course, we'll need to figure out how to handle that on earlier seasons where scrambles were not specifically marked in the play-by-play).
|Rk||with Pass Pressure||without Pass Pressure||Difference|
|Rk||with Pass Pressure||without Pass Pressure||Difference|
Note: This table appears as "2013 Quarterbacks with and without Pass Pressure" in the Statistical Appendix of FOA 2014, but the DVOA numbers are incorrect. In the original version of the book, this table accidentally used the DVOA version with offense as the opponent adjustment instead of defense. The correct numbers are posted above here, and those of you who have purchased the book in PDF form can now download a corrected version, free of charge. We apologize for any inconvenience.
The teams who used multiple quarterbacks are always fun to study, but they really stand out here. Jason Campbell actually had the smallest pressure drop for DVOA (-41.6%) in the last four seasons. He's the third unspectacular quarterback to lead the league after Shaun Hill (2010; -42.4%) and Josh Freeman (2011; -65.3%). Campbell somehow ranked third with pressure and was the fifth lowest without pressure. Brandon Weeden had a much different experience behind Cleveland's offensive line. He was the second worst with pressure and still well below average without.
If you don't pressure Aaron Rodgers, there's really no hope for your defense. He ranked third in DVOA without pressure, but fell to 34th with pressure, including a broken collarbone on a sack against the Bears. That eventually led to the return of Matt Flynn in Green Bay, and no one fared worse against pressure. Rodgers and Flynn had nearly identical pressure rates and both averaged 7.4 fewer yards per play under pressure -- the highest declines in the league.
Marc Trestman's Chicago Bears may have the most mystifying results of any 2013 offense. Josh McCown had the highest DVOA (8.0%) with pressure. He's the only qualified quarterback in the last four years with a positive DVOA under pressure. Jay Cutler finished a respectable second, but the difference really came without pressure. Cutler fell to 25th while McCown was still strong at seventh. I'm certainly not going to say McCown will carry this success to Tampa Bay, but he made one of the best plays last year under pressure: a touchdown pass to Brandon Marshall on third down in the first quarter of the Week 9 win over Green Bay.
Inexperienced passers Terrelle Pryor and Case Keenum were two of the three most pressured quarterbacks, but they brought a lot of that on themselves. Notice that Matt Schaub had a pressure rate nearly 10 percentage points lower than Keenum, and Matt McGloin was 12 percentage points lower than Pryor. Those were the two biggest pressure rate differences among our eight teams with multiple quarterbacks. The third biggest was Minnesota with Matt Cassel handling pressure much better than Christian Ponder. However, only Pryor had a lower DVOA without pressure than Cassel. Realistically, how long can the Vikings go without giving the keys to Teddy Bridgewater?
I've noticed a lot of negative talk this week about Buffalo's EJ Manuel. As a rookie, he had the third-lowest DVOA without pressure (10.3%). Granted, our data only goes back to 2010, but this is a very discouraging sign. We have 40 examples where a quarterback had a DVOA without pressure under 20.0%. Not one of those quarterbacks was a good starter after that point. Now Eli Manning (14.9% in 2013) has a great shot to rebound, but where he's at entering his 11th season is way beyond where Manuel is at. Matt Schaub (15.4% in DVOA) might get back on track in Oakland, but who would want to bet on that? As for Sam Bradford, let's just say we're still waiting to see a good quarterback there. It's not a good sign if your quarterback's not taking advantage of adequate protection. Even Thad Lewis (not enough attempts to rank) had a 31.5% DVOA without pressure in Buffalo last year. However, that leaves Lewis with the biggest pressure drop (-232.0%) in four years. Buffalo has some issues at quarterback. What else is new?
Robert Griffin III (27th) and Matt Ryan (12th) were the only quarterbacks to have the same rank in DVOA with and without pressure. Some may have expected Ryan Tannehill to have a huge difference in his rankings, and to some extent, he did. Few would have expected his ranking of 13th in pressure rate. Yes, Tannehill faced a below-average amount of pressure in 2013. Tannehill's problem was that pressure really made him struggle, ranking 38th in DVOA. He was much more respectable without pressure at 15th, but he has to handle pressure better. Most of his league-leading 58 sacks were on the offensive line, but he still suffered the fourth-biggest pressure drop in the league. The split was a little similar in his rookie year when he ranked an unimpressive 26th in pressure drop.
Philip Rivers was dead last in pressure drop (-211.2%) in 2012. Mike McCoy and Ken Whisenhunt put better blocking in front of him, got the ball out of his hands faster, and Rivers moved up to 18th in pressure drop last year. Rivers is another intriguing case, because he has two of the top 12 seasons since 2010 for DVOA with pressure (2010 and 2013). Rivers (two) and Roethlisberger (three) are the only quarterbacks with multiple seasons in the top 25 for DVOA with pressure. In 2013, Rivers was one of six quarterbacks to rank in the top 10 for DVOA with and without pressure (the others being Nick Foles, Peyton Manning, Russell Wilson, Josh McCown and Colin Kaepernick).
[ad placeholder 3]
Generally, the best quarterbacks in the league will play like the best quarterbacks if you don't get pressure on them. Last year, that familiar group of Peyton (second), Rodgers (third), Brees (fifth) and Brady (10th) all ranked in the top 10 in DVOA without pressure. What about other veterans? I used everyone with at least three qualified seasons (minimum 200 plays) since 2010 and ran a few year-to-year correlation tests. Peyton and Chad Henne are missing 2011 in their data, but that's okay since they're actually two of the most consistent quarterbacks (just in opposite directions of success). Our sample contains 97 seasons by 27 quarterbacks, so it's not the most statistically sound test, but it's a solid start.
|DVOA with pressure||0.18|
|DVOA without pressure||0.58|
Similar to studies on sacks, we see that quarterbacks have a considerable impact on controlling how much pressure they face. If you're getting rid of the ball quickly, then it's hard for the defense to get that close. Dance around and even the best offensive linemen will struggle to hold their blocks. We know linemen come and go, so pressure rate's 0.44 correlation is a solid indicator of the quarterback's impact. The 0.58 correlation for DVOA without pressure is even better. Quarterbacks can play at a fairly consistent level each year when the element of pressure is gone and it's just a matter of making reads and throws. Add pressure, and you're going to get a lot of variation in performance, hence low correlation (0.18). Even if you're a great quarterback, it's almost impossible to sustain a high level of play under pressure.
That's why we remember the highlights best (or lowlights for players perceived to be bad under pressure). That's why I'm still in awe over McCown's touchdown pass in Green Bay, but struggle to remember similar plays from him last year. The successes are rare. Would McCown have a positive DVOA under pressure if he played a full season instead of 245 plays? Our data suggests he wouldn't, but the great part is that we actually have this data now, and will continue it for each season going forward.
Unsurprisingly, pressure drop had a very small, albeit non-zero correlation (0.02). Pressure's randomizing effect on quarterback performance makes it too hard to expect a consistent difference between plays with and without pressure.
Quarterbacks are only getting pressured on about a quarter of all dropbacks, but those are the plays that can really make the difference in the game. Get a third-down stop here, or force a pick there. Pressure is a defense's best weapon.
Nothing represented this better in 2013 than the Denver Broncos and Seattle Seahawks. Peyton Manning had the lowest pressure rate in the league for the second year in a row. Seattle's offense had the worst pressure rate with Russell Wilson ranking 39th individually. Yet in Super Bowl XLVIII, Wilson was not sacked or even hit once. Manning trailed 15-0 after throwing one incompletion, which was a bad interception after Cliff Avril brought some pressure from the edge. Manning's second interception was also caused by Avril, who hit the quarterback as he threw. The ball hung in the air and Malcolm Smith returned it 69 yards for a touchdown.
Denver trailed 22-0, the game was practically over, and the MVP quarterback had thrown only two incompletions of consequence. Both were the result of third-down pressure. Seattle may not have protected its quarterback well for the first 18 games of the season, but it also had the league's best defense at putting pressure on the opposing quarterback. Two plays were all Seattle needed to embarrass a record-setting offense. Of course, Denver only needed the opening snap to embarrass itself, but the super blowout was paced by a couple of pass pressures.
In part two later this week, we'll look at Seattle and the other 31 defenses' success with and without pass pressure.
27 comments, Last at 15 Aug 2014, 11:56pm
#1 by DisplacedPackerFan // Aug 11, 2014 - 1:50pm
I think part of the issue for Green Bay and very poor numbers under pressure last year, when I recall DVOA being much better in prior years, was the receivers. Driver, Jennings, Cobb, and Finley were all very good on broken plays in previous years. Nelson, as great as he is, hasn't ever stood out to me on broken plays. He's had some great plays on roll outs and some of his sideline grabs have come when Rodgers has been fleeing, but those results feel more like Rodgers/Flynn getting him the ball at the proper time despite the pressure than him helping Rodgers/Flynn out. Boykin and Jones are both just solid #2/3 guys that GB can make into 1,000 yard receivers if they play well for a full season so you don't expect much from them either.
The Driver/Jennings/Finley/Cobb type guys would clearly break a route to make plays when pressure hit. All of them were gone for all or most of last season, or longer in the case of Driver. Cobb and Finley each played 6 games last year and not the same 6 so maybe you could see if there was a big difference in DVOA with pressure in the games they did play vs the ones they didn't. I expect that the Packers QB's, and hopefully it's just Rodgers, will look better under pressure again this year if Cobb is healthy all season. Richard Rodgers at TE might turn out to be a guy that can help in those situations as well, but tough to say anything about rookies.
In general I really do wonder about the effect of receivers on how a QB does under pressure. Sticking with the NFC North, the teams I see the most, Jefferies and Marshall seem like two great guys to have out running routes when things break down, as does Matt Forte. You'd think Calvin Johnson and Reggie Bush would help any QB when under pressure too. Jennings may have helped Cassel look a bit better than he otherwise might have.
I'm not saying the QB doesn't have a huge part in it, but a Gronk-less/Welker-less Brady wasn't great, but I bet he looked better with one/both of them this year/previous years.
Not sure how you would quantify a "pressure" receiver, and sample sizes get even smaller, but number or targets when a QB is under pressure, and offensive DVOA when those guys are targeted under pressure vs the rest of the team might show something.
#5 by DisplacedPackerFan // Aug 11, 2014 - 4:00pm
It doesn't, I was mostly just calling Jones and Boykin JAGs (just a guy). They don't excel at anything, but they don't really suck at anything. That includes doing well on broken plays. I think what made them stand out at times had more to do with the QB. Boykin can still improve, he's young. Jones I expect is never going to be special. He may be a 1000 yard receiver at some point, but in today's NFL that isn't much more than JAG level.
Being a #1 doesn't make you good at it either. Nelson is a #1 receiver but I've never thought he was good on broken plays.
I think practice on it matters too. The Packers have such awful coverage linebackers that they will never get good practice in dump off passes, which is important on broken/pressure plays. It's not where the big DVOA will come from but it matters.
#6 by theslothook // Aug 11, 2014 - 5:24pm
in all the years sproles has been on thr team, never did i think the saints had good cover lbs. Onr would think practicing against a great o or d would make the other side better, but it never seems to play out this way
#18 by KB // Aug 13, 2014 - 11:39am
I personally see Nelson a bit different than you. Cobb is amazing when the play breaks down but imo Nelson isnt that far behind. There was an article about this after he signed his contract where Rodgers touched on this subject. I cant find it but if I remember right he was basically saying it takes a smart wr but just as important is chemistry. Outside of what writers and teammates say id be very curious to see advanced stats on this subject. Id personally think Nelson would be the better option and probably would fare better stastically in this situation than jeffery or marshall because of how smart he is and the fact that him and Rodgers have one of the best connections among any qb wr duo. Their back shoulder throw is the best in the nfl and a thing of beauty. Easily one of my favorite plays in the nfl. I really hope fo or pff dive into this subject if they havent already.
#19 by Perfundle // Aug 13, 2014 - 1:27pm
Chemistry on back-shoulder throws is quite different from chemistry on scramble plays; the first is by design and the second is completely improvised. I'm not sure if there's even any chemistry to develop, because the idea is for receivers to look back after running their route for a bit, and if their QB is scrambling then run in the same direction he's running in, trying to give him a clear passing lane to you.
#24 by LionInAZ // Aug 14, 2014 - 7:49pm
I recall tuluse once commenting that Packer WRs are picked for the specific ability to not look for a pass until after it's thrown. If that's true, it would hinder their ability to help out the QB under quick pressure. Receivers running over the middle (like say, Cobb or Driver) might see the pressure and bail out. My impression has been that Nelson doesn't run routes over the middle much, so he's not in a position to bail out Rodgers very often.
#26 by KB // Aug 15, 2014 - 1:18am
Nelson played primarily in the slot last year and a lot of snaps came over the middle. Nelson as a wr really can do it all at an elite level. If I wasnt on my phone id look up the % of catches he had over the middle last season.
I completely disagree about chemistry when plays break down because of pressure. Each QB is different and will want things done a certain way which in reality is with anything between those two positions though.. Knowing where and when your wr will be as a qb would help exponentially. Teams even practice such drills. If there wasnt chemistry to build then what would really be the point. Imo its a lot more than a wr breaking off his route and running a random direction.
#3 by theslothook // Aug 11, 2014 - 3:15pm
I think the real highlight of this article was the fact that pressure is absolutely a random variable and we have no idea what the "true" value for these qbs is as of now. Maybe with 5-6 years of charting data, we will start to unearth a growth pattern with these younger qbs, but as of now, we're still in the dark.
#4 by Scott Kacsmar // Aug 11, 2014 - 3:26pm
Just for some receiver context.
Most dropped passes from a QB under pressure:
Jay Cutler - 8 (McCown had 4 so the CHI numbers could have been even better)
Russell Wilson, Ryan Tannehill, Terrelle Pryor - 6
Andrew Luck, Matt Ryan, Brandon Weeden, Geno Smith, Mike Glennon - 5
Most air yards lost on a dropped pass while QB was under pressure:
#7 by nickbradley // Aug 11, 2014 - 6:48pm
This could have been a lot cooler. DYAR pressure/no pressure splits would be a good idea. I Can't do that, but I did a bit of an adjusted DVOA, and what I did is look at what a QBs DVOA would be if they were pressured at the league-average rate of 25%. =[25%]*[pressure DVOA] + [75%]*[no pressure DVOA].
first thing that stands out is how well Wilson did for being pressured 37% of the time. Here's the list:
Qualifying QB Plays Pct adj DVOA
N.Foles 366 0.26 59%
J.McCown 245 0.26 51%
P.Manning 688 0.15 49%
P.Rivers 601 0.22 49%
R.Wilson 508 0.37 44%
C.Kaepernick 507 0.26 38%
D.Brees 703 0.23 34%
A.Rodgers 330 0.22 34%
T.Brady 680 0.20 22%
M.Ryan 712 0.29 21%
B.Roethlisberger 639 0.22 20%
J.Cutler 386 0.26 19%
A.Luck 662 0.27 18%
C.Newton 557 0.24 18%
T.Romo 585 0.22 14%
R.Fitzpatrick 398 0.23 14%
J.Locker 214 0.27 13%
C.Palmer 632 0.27 13%
M.Glennon 480 0.32 12%
K.Clemens 274 0.29 9%
A.Smith 601 0.21 8%
M.Stafford 671 0.19 7%
J.Campbell 346 0.25 4%
C.Keenum 276 0.38 2%
R.Griffin 533 0.28 2%
A.Dalton 645 0.16 0%
C.Ponder 296 0.28 0%
S.Bradford 289 0.25 0%
R.Tannehill 659 0.22 0%
G.Smith 521 0.29 -2%
M.McGloin 224 0.22 -3%
J.Flacco 690 0.25 -5%
M.Cassel 283 0.21 -8%
E.Manuel 369 0.22 -10%
M.Schaub 388 0.28 -10%
C.Henne 556 0.24 -11%
T.Pryor 338 0.34 -11%
E.Manning 607 0.28 -13%
B.Weeden 301 0.28 -19%
M.Flynn 239 0.22 -23%
#8 by Karl Cuba // Aug 11, 2014 - 8:35pm
It's a nice idea and useful as a rough effort but it doesn't take account of how much responsibility a qb carries for the pressure in the first place (though I'm not sure how much blame Wilson carries behind that line).
In a perfect world you'd be able to include data on how long the pressure took to arrive.
#12 by nickbradley // Aug 12, 2014 - 11:55am
Well yes -- but I don't think this analysis assigns responsibility between QB and OL either. I don't really have a problem treating the quarterback and his pass blockers as the same entity.
** Something else I found interesting is that the bigger the pass/run DVOA split is for a team, the *less they are pressured...seems backwards.
r^2 is just .121, but the p-value are quite tiny -- 1.76464E-29 for the constant and .0275 for the relative DVOA difference.
#10 by Peregrine // Aug 12, 2014 - 10:19am
Thanks. I took the data and calculated the number of plays under pressure and to this Falcon fan's complete lack of surprise, Matt Ryan was pressured most with 204 pressure plays. (Russell Wilson was second with 186, and credit to him for doing so well both pressured and not.) And I would say that very early on the coaches realized pass protection was a catastrophe and emphasized play calls that got the ball out quickly, which Ryan can do well. Even then, Ryan got pounded all year.
Maybe with better pass protection, a better running game, heathier skill weapons, and a better defense helping on the other end, Ryan won't get the hell beat out of him this season. It's no accident that the theme on Hard Knocks is all about toughness. Here's hoping.
#11 by MJK // Aug 12, 2014 - 11:47am
Unless I'm misunderstanding, this article is talking about how QB's respond when the defense is SUCCESSFUL in bringing pressure, which is perhaps the less interesting thing to study. As the intro says, every QB fares worse under pressure, so a defense always wants to achieve that as possible.
The more interesting thing is how a QB does when the defense TRIES to create pressure (i.e. blitzing, or using creative rushes) above and beyond their "normal" mode. To wit, I would be curious to see a similar analysis of how a QB does versus "blitzs and creative rushes" compared to "normal" (rush 4) and "coverage" (rush 1-3) scenarios. I'm sure such data exists--I think FO has run some articles on it in the past. Plus, it's a lot less ambiguous than trying to define "pressure".
After all, a QB has some control over whether the pressure succeeds or not, since he helps set protections and can audible to a hot read if he expects pressure. That should be part of the evaluation.
#13 by Scott Kacsmar // Aug 12, 2014 - 2:06pm
I don't like "QB vs. blitz" stats because if the pressure is picked up, then the QB should have an easy time completing a pass. Maybe we can look at DVOA vs. an unsuccessful blitz next year, but I would imagine the numbers are great. Greater than DVOA without pressure? That would be interesting to see, but generally I couldn't care less about a "vs. blitz" column.
#14 by Perfundle // Aug 12, 2014 - 3:18pm
I don't think that would be particularly interesting, because the best QBs against the blitz are generally the best QBs overall; if that wasn't the case, all defenses would have to do is to blitz those QBs and they wouldn't be the best QBs anymore.
#15 by Scott Kacsmar // Aug 12, 2014 - 3:50pm
I'm interested in the league-average numbers to see if my theory about a failed blitz being easy to beat holds up. And to see if a good blitz is harder than a regular four-man pressure.
So it would be...
League-average DVOA vs. failed blitz
League-average DVOA without pressure (non-blitz plays)
League-average DVOA vs. successful blitz
League-average DVOA with pressure (non-blitz plays)
Just seeing completion percentage and YPA would be nice too. I would imagine a lot of passes against the failed blitz are still coming out quickly, so maybe they don't gain as many yards.
#16 by ayjaymay // Aug 12, 2014 - 4:07pm
Analysts like to point to the Giants' offensive line woes in 2013 to explain Eli Manning's putrid play last year. This shows that he was horrible whether he was pressured or not. He wasn't pressured significantly more than many other quarterbacks, whom all performed much better. Russell Wilson really stands out as one of the most pressured QBs in the league (patchwork offensive line for more than half the season), yet was able to perform as one of the best. I'm surprised that Big Ben wasn't pressured more, given the Steelers' OL troubles last year.
#22 by theslothook // Aug 14, 2014 - 3:12pm
I'm not sure it's that well known. The obvious one's are Brady Manning and Brees. Rodgers to some extent but he also holds the ball. We know others like Matt Ryan probably do as well. But to what degree or what extent is more up in the air. And what about people like Stafford, Cutler, and Romo?