Quarterbacks and Pressure, 2015
by Carl Yedor
In light of David Bowie's passing in January, it would make sense to open with a line from the classic Under Pressure as we begin our discussion of how quarterbacks fared when under duress from the opposing defense. However, Scott Kacsmar did the same two years ago, so let's keep it simple and say that in 2015, just as in every other year of tracking data we have, quarterbacks with a qualifying number of plays were clearly worse off when faced with pressure. Today, we'll take a look at the offensive pressure numbers from the charting results, which will appear in Football Outsiders Almanac 2016 (currently scheduled to be released sometime between July 26 and August 1).
A quick note first on where this year's pressure numbers come from. In 2015, we got charting from two sources, ESPN Stats & Info and Sports Information Solutions. Each company charted pressure somewhat differently. Various time delays this offseason prevented us from fully checking all plays where the two companies disagreed on pressure, but we did check a lot of them to get a more accurate picture of pass pressure. Overall, our pressure stats are closer to those of ESPN, but informed by the charting done by SIS.
Interestingly, in a change from Sterling Xie's work from last year, the correlation coefficient between pressure rate and pressure DVOA in 2015 was 0.26, as opposed to -0.16 in the 2014 season. Instead of an increased pressure rate leading to worse DVOA when pressured, the result flipped. This is still not a very strong relationship, though, and it suggests that there is not a consistent relationship between how frequently a quarterback faces pressure and how adept he is at handling it.
Before we go into the numbers, it is important to note that DVOA listed here is not the individual quarterback's passing DVOA. Instead, it is team offense DVOA with this quarterback either passing (including sacks) or scrambling.
You might have been able to guess this from his playoff performance against the Chiefs, but Brian Hoyer had the worst drop-off in DVOA on all passes and scrambles from non-pressured plays to pressured ones. On plays without pressure, his DVOA was 45.0%. As soon as pressure was introduced into the equation, that number fell all the way to -144.9%, for a difference of -190.0%. (Rounding the decimals accounts for the missing 0.1%.) Fortunately for the Texans, out of 37 qualifying quarterbacks, Hoyer had the seventh-lowest pressure rate, so they were still able to win the AFC South in spite of his struggles with men in his face.
On the other end of the spectrum, Jay Cutler had the smallest drop in performance when faced with pressure, with a difference of "only" -45.0%. Cutler posted a -4.3% DVOA against pressure in 2015, the fifth-best DVOA against pressure in our database going back to 2010. Two of those seasons belong to Ben Roethlisberger (2010, 2014), and the other two belong to Josh Freeman (2010) and Josh McCown (2013). There is a 50-50 chance that you get a good quarterback in that list. I think that's appropriate for Cutler, because half the time he makes you want to pull your hair out and half the time he pulls a third-down conversion out of thin air.
It was a good thing, too, as Cutler faced the eighth-most pressure, behind Aaron Rodgers, Brock Osweiler, Russell Wilson, Teddy Bridgewater, Colin Kaepernick, and the Cleveland quarterbacks. Johnny Manziel and Josh McCown were nothing to write home about, but their offensive line certainly did not do them many favors in pass protection.
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.
|Qualifying Quarterbacks with and without Pass Pressure, 2015|
A star in this exercise from previous years, Ben Roethlisberger was in the top half of quarterbacks in terms of smallest DVOA difference again. However, that was with a difference of -102.0% and a DVOA when under pressure of -41.0%. Big Ben only faced pressure on 18.9 percent of his pass plays, so the drop-off in team offensive DVOA against pressure did not hurt as much.
Russell Wilson, another quarterback known for extending plays with his scrambling ability, posted a DVOA difference in the bottom half of the league, but that was largely because he finished with the best DVOA without pressure at 86.4%. His DVOA with pressure of -32.4% was the fourth-best of all qualifying quarterbacks behind Cutler, Carson Palmer, and Ryan Fitzpatrick. It's a good thing for NFC defenses that the Seahawks' offensive line was so dreadful in 2015, allowing pressure on 31.7 percent of all passes and scrambles.
Over the past three years, Wilson has finished third, fourth, and second in knockdown rate. Part of that is likely due to his tendency to hold onto the ball and scramble, but that is something of which the team should be cognizant.
[ad placeholder 3]
Jameis Winston was knocked to the ground more than any other quarterback in 2015. He broke the three-year streak of another former first-overall draft pick, Andrew Luck, who led the league in knockdowns in 2012, 2013, and 2014. Last year, Luck only played in seven games, while maintaining the sixth-highest knockdown rate in the league. Winston, meanwhile, finished 10th in the league in DVOA against pressure, which is encouraging when compared to the rookie year struggles of 2014's highest-drafted quarterback, Blake Bortles (third-worst). Winston's first-round rookie compatriot Marcus Mariota was knocked down less frequently than Winston (16.0 percent versus 20.2 percent) while spending more time in the pocket (16.7 percent of pass plays/scrambles outside the pocket versus 18.5 percent for Winston). However, Winston was much better in the face of those knockdowns, with Mariota posting the fifth-worst DVOA against pressure in the league.
Somewhat surprisingly, given that he ended up getting benched for Brock Osweiler, Peyton Manning had the fourth-smallest drop in DVOA when pressure was added to the equation. A closer look shows that this was mostly due Denver's team offense DVOA being only 5.5% without pressure, compared to -74.0% with it. Manning's mark with pressure was slightly below average in 2015, but that 5.5% mark without pressure is the worst among quarterbacks with at least 200 plays since 2012, when the triumvirate of Matt Cassel, John Skelton and Brady Quinn all posted a worse DVOA without pressure. I sure never thought I would be making a reasonable statistical comparison between Manning and this guy:
For those wondering, Osweiler's numbers from 2015 were both right about average. Given all that, it's probably for the best that the Sheriff rode off into the sunset after Denver's Super Bowl victory. His counterpart at Levi's Stadium that day, Cam Newton, finished solidly above average in DVOA both with and without pressure on the season, including an improvement of 44.2% when facing pressure from 2014 to 2015. This may be partially due to Carolina facing the weakest schedule in the league by DVOA in 2015, but you can only play the team that lines up across from you.
It seems like there is a rule that a Bengals game cannot go on the air without the discrepancy in Andy Dalton's performance with and without pressure being mentioned. The way color commentators have told it in the past, Dalton is a solid quarterback when not facing pressure, but as soon as defenses turn up the heat, he turns into a piñata waiting for rushers to hit him and collect the turnovers that spill out.
Dalton has improved substantially when faced with pressure over the past few seasons, raising his DVOA mark from -117.8% in 2013 to -92.5% in 2014 and -51.4% in 2015, good for 11th in the league last year. His drop-off when pressure was added was still quite large in 2015 at -129.9%, but much like Russell Wilson, this was due to the fact that he posted a tremendous DVOA (78.6%) sans pressure. Fortunately for Cincinnati, Dalton faced pressure on a league-low 18.5% of his passes and scrambles, allowing Dalton to shine more often than not.
Dalton's improvement against pressure of 41.2% from 2014 to 2015 was the fifth-largest year-over-year change behind Newton (44.2%), Cutler (65.1%), Blake Bortles (83.3%), and McCown (118.6%). At first glance, this would appear to be a great sign for the Jaguars, given how much Bortles improved between his rookie and sophomore year. However, the five largest declines from 2014 to 2015 belonged to Roethlisberger (-40.0%), Kaepernick (-41.2%), Hoyer (-43.6%), Philip Rivers (-46.5%), and Nick Foles. In past years, Roethlisberger has been one of the best by this measure, so it appears that any huge jumps and falls are more likely due to simple regression toward the mean.
[ad placeholder 4]
In 2014, Bortles and McCown were both in the bottom three quarterbacks against pressure (Robert Griffin III was the other). As a result, they had basically nowhere to go but up. Now, there are certainly other factors that would suggest some of Bortles's improvement could stick, because he also improved his DVOA without pressure from an abysmal 6.5% in 2014 to 34.1% in 2015. DVOA without pressure is often a better indicator than overall DVOA of how good a quarterback is, and given that Bortles is entering his age-24 season, he still has a lot of room for growth when not facing pressure. I wouldn't bet the house on all of that improvement against pressure sticking, but even if it doesn't, the AFC South is still wide open, meaning that Jacksonville could win the division even with some regression from their quarterback while under pressure.
McCown presents a particularly interesting case, in that his DVOA when under pressure fell 163.6% from 2013 to 2014, only for him to bounce back with an improvement of 118.6%. His DVOA without pressure went from 65.2% in 2013 to 32.8% in 2014 to 31.4% in 2015, compared to yo-yoing from 8.0% in 2013 to -155.6% in 2014 to -37.0% in 2015 with pressure. The presence of Robert Griffin III on the roster in Cleveland may prevent us from seeing whether McCown's performance under pressure would continue to jump around in 2016.
Griffin's replacement in Washington, Kirk Cousins, had a sharp drop in performance when pressure was introduced, falling from a DVOA of 69.0% without pressure to -110.1% with it. Opposing teams only managed to get pressure on Cousins 20.5 percent of the time, the fifth-lowest rate in football, which certainly helped him put up a career year. Washington's weak schedule (25th-ranked by DVOA) also played a part in his drastic improvement, so it will be interesting to see whether Cousins can keep up his level of play against tougher competition.
Later this week, we'll be looking at pressure from the defensive side of the ball and how successful teams were at getting after the quarterback.
4 comments, Last at 26 Sep 2017, 9:34pm
#1 by Noahrk // Jul 13, 2016 - 12:51pm
What's the effect of stuff like bubble screens on the numbers, I wonder? I keep thinking back to the insane number of failed bubble screens Miami's inept offense kept calling. Those plays would chart as non-pressure (and in Miami's case) low-DVOA plays. But were there enough of them to strongly skew the numbers? Typically a QB would bear part of the responsibility for not audibling out of a bad play, of course, but Miami's offense had no audibles. Naturally.
#2 by nosoop4u // Jul 14, 2016 - 1:51pm
Thought it would be interesting to see how the QBs would be rate if they faced league average pressure (25.86%), so I weighted them as if they did (DVOAp x .2586 + DVOAnp x (1-.2586)), and here's how they came out. Gives us an interesting theoretical where the OL influence is reduced in quantifying a QB's performance. Not sure if DVOA is additive this way, so the resulting numbers may better be called something else...