Football Outsiders

Innovative Statistics, Intelligent Analysis

2017 Play-Action Defense

by Bryan Knowles

Earlier this week, we looked at play-action offense around the league. It was the in thing in 2017, rising to levels we've never before recorded. Today, we flip to the other side of the ball and check out how defenses handled the increase.

Sometimes, one team breaks a stat. Horrendously. When you're dealing with only 32 data points, one team that breaks everything can sometimes unfairly skew the numbers. Last year, we puzzled at why there was no correlation from season to season in how well a team defended play-action. There had been correlation in the years prior, but it vanished in 2016. It also remained gone this year -- there was just a .008 correlation coefficient between a team's defensive play-action DVOA in 2016 and their DVOA in 2017. That looks like just random noise, but that's the fault of exactly one team: the 2016 Houston Texans.

Houston saw a 92.6% drop in their DVOA, going from -31.5% in 2016 to 61.1% last season when defending play-action passing, which is insane. In 2015, they had a 53.9% DVOA. If you toss out the 2016 Texans and just look at everyone else, there's a much more reasonable .35 correlation between last year's results and this year's results. That makes a ton more sense.

That's not to say that there are not dramatic year-to-year swings. Even excluding Houston, eight teams saw their DVOA get worse by at least 15 points last season. On the flip slide, Carolina saw their DVOA improve dramatically, going from 23.2% to -42.7%, as one of nine teams to see their DVOA increase by at least 15 points. Small sample size does lead to large swings, but even by that standard, this is somewhat ridiculous.

In the following table, 2017's defenses are sorted by how frequently they faced play-action passes. They are also listed with numbers against play-action passes (including sacks and scrambles) and passes without play-action. The final column shows the difference between play-action and regular dropbacks, with a lower number there indicating a defense that was better against play-action.


Play-Action Defense, 2017
Defense PA Pct Rank With PA (Pass/Scram) No PA Difference
Yds/Play DVOA Rank Yds/Play DVOA Rank Yds/Play DVOA Rank
NYG 27% 1 7.1 4.6% 9 7.2 19.1% 31 0.0 -14.5% 4
CLE 26% 2 8.0 44.5% 29 6.5 16.6% 21 1.5 27.9% 25
IND 26% 3 9.3 41.1% 28 6.8 26.2% 30 2.5 14.9% 17
LAC 24% 4 6.9 19.2% 17 5.4 -12.6% 5 1.5 31.8% 26
NYJ 24% 5 7.5 32.3% 22 6.5 14.4% 23 1.0 17.9% 19
SEA 23% 6 6.9 12.8% 13 5.4 -0.7% 7 1.4 13.5% 15
ARI 23% 7 8.7 40.4% 26 5.0 -18.8% 2 3.7 59.2% 32
SF 23% 8 7.1 14.0% 14 6.5 26.3% 25 0.6 -12.3% 6
OAK 23% 9 8.0 16.5% 15 6.6 30.0% 26 1.4 -13.5% 5
GB 22% 10 8.8 40.6% 27 6.5 18.5% 24 2.3 22.1% 21
TB 22% 11 7.5 9.4% 12 7.3 33.7% 32 0.2 -24.3% 2
BUF 22% 12 6.7 3.7% 8 6.1 1.2% 17 0.7 2.5% 11
KC 22% 13 7.6 26.6% 19 6.7 19.4% 29 0.8 7.2% 14
DAL 22% 14 6.5 17.1% 16 5.8 12.8% 13 0.6 4.2% 12
HOU 22% 15 10.3 61.1% 32 6.5 5.8% 22 3.9 55.3% 30
MIN 22% 16 5.7 -14.1% 2 5.3 -11.6% 4 0.5 -2.5% 9
Defense PA Pct Rank With PA (Pass/Scram) No PA Difference
Yds/Play DVOA Rank Yds/Play DVOA Rank Yds/Play DVOA Rank
MIA 22% 17 7.8 44.9% 30 6.2 18.8% 18 1.6 26.1% 24
PIT 22% 18 9.1 33.3% 23 5.1 -16.8% 3 4.0 50.1% 28
DET 21% 19 6.6 5.9% 10 6.7 11.9% 28 0.0 -6.0% 7
CIN 21% 20 5.4 -1.4% 6 6.0 15.1% 15 -0.6 -16.4% 3
WAS 21% 21 6.4 -3.6% 5 6.0 -8.5% 16 0.4 4.9% 13
CAR 20% 22 4.9 -42.7% 1 6.6 8.3% 27 -1.8 -51.0% 1
BAL 20% 23 7.2 31.0% 20 5.4 -25.5% 6 1.7 56.5% 31
LAR 20% 24 6.5 0.4% 7 5.8 -14.4% 10 0.7 14.8% 16
NO 20% 25 6.5 -8.2% 3 6.3 -5.4% 20 0.1 -2.8% 8
PHI 20% 26 7.4 8.0% 11 5.6 -7.6% 8 1.9 15.6% 18
ATL 20% 27 7.3 34.8% 24 5.8 14.5% 14 1.5 20.2% 20
CHI 19% 28 7.4 32.1% 21 5.8 -1.1% 11 1.7 33.1% 27
TEN 19% 29 6.9 20.3% 18 5.8 18.5% 12 1.1 1.8% 10
NE 19% 30 8.1 37.1% 25 6.3 13.9% 19 1.8 23.2% 22
DEN 18% 31 8.1 51.4% 31 5.7 -1.2% 9 2.3 52.7% 29
JAX 17% 32 6.9 -5.0% 4 4.5 -29.2% 1 2.3 24.2% 23
NFL 22%
7.3 18.9%
7.4 18.2%
1.3 13.7%

Who are the best pass-covering linebackers in the league? Any list of the top 10 or 15 would undoubtedly include Thomas Davis, Luke Kuechly, Eric Kendricks, and Anthony Barr. Perhaps, then, it's not surprising that Carolina and Minnesota ranked one-two in best play-action pass defenses; that's a lot of quality defending going on out there. You're not going to fool those four with your fake hand-offs and your trickery; they can smell a play-fake coming from a mile away.

It's worth diving deeper into Carolina because, wow, -42.7%? That's incredible. It's the best we've recorded since 2009, and by a substantial margin. It's also, as we mentioned, the biggest improvement we saw last season. Carolina ranked 17th in play-action defense in 2016, purely middle-of-the-pack. And now they vault to record-setting success? That is surprising.

Last year, the Panthers gave up 8.5 yards per play on play-action, and that plummeted to a league-best 4.9 last season. My first thought when seeing that was "oh, teams were throwing shorter play-action passes against Carolina last season, for whatever reason, resulting in a lower DVOA; it's the Philadelphia RPO thing all over again." But no, that's not the case; opponents had an average depth of target of 8.4 yards when passing off play-action against Carolina. That's slightly below average, but nowhere near the bottom. Heck, Atlanta faced an average target depth of just 6.8 yards, and they really struggled defending play-action, so that's not it.

Nor did Carolina generate a ton of pressure on play-action passes. They had a 35.7 percent pressure rate on play-action plays, which is again only slightly above average. They didn't force opponents to scramble more frequently than others or intercept more passes or anything of that nature. They did allow a completion rate of just 57.7 percent, seventh-best in the league, so that's something, but that's really why their DVOA was so great.

The one thing that really jumps out is how well Carolina did when they allowed a completion. They had a defensive DVOA of 68.9% when their opponents successfully completed a play-action pass. The average team had a DVOA of 134.6%, and only two other clubs were below 100.0%. The average complete play-action pass against Carolina traveled just 4 yards through the air, and the 5.9 YAC they allowed was below average, as well.

Unlike with Philadelphia, this wasn't an offensive strategy at work; Carolina didn't face an unusually high number of screens or quick slants, and their overall depth of target was within a yard of the league average. For whatever the faults of Carolina's pass defense in 2017 -- and there were plenty -- they did a good job not biting on play-action, sticking with their man and forcing offenses to check down more often than not, and then they did a good job swarming to the ball and preventing massive gains once a catch was made.

Not biting on play-action and preventing the big gain is a really good way to improve your play-action DVOA. Carolina, Cincinnati, and New Orleans had the shortest aDOT on completed passes; they're all in the top six defenses defending play-action. The bottom three in aDOT were Pittsburgh, Indianapolis, and Houston, and yup, they're all in the bottom ten in play-action defense as well.

But let's get back to Houston, the team that single-handedly wrecked year-to-year correlation. The only defense to be worse than their 61.1% play-action DVOA in the past decade was 2015 New Orleans, and when you're being compared to the Saints defenses of the 2010s, you know you're in some shoddy company. Houston gave up a 65 percent completion rate on play-action, and those passes traveled an average of 9.5 yards in the air, longest in the league. To a certain extent, you can get away with either allowing a ton of completions (Oakland allowed a 70 percent completion rate) or having those completions be deep downfield (the average play-action completion against Jacksonville travelled 7.7 yards downfield, compared to an NFL average of 6.6), but not both.

And remember, Houston led the league in defensive play-action DVOA last season!. We gave a lot of credit to Brian Cushing returning to form and Benardrick McKinney taking a step forward, but it's looking more and more like it was just a weird one-year fluke. Houston was dead last in 2013, and has only had one negative DVOA season against play-action since 2011. How this team ranked first in 2016 is fairly mysterious.

If Houston had been good for years and years, and then dropped off with the rest of the defense last season, then that would make sense -- the loss of Whitney Mercilus could be something you could point to for the decline and a reason to expect improvement going forward, for instance. But no, when a team has back-to-back seasons with 80-point swings in their DVOA, you just have to sort of frown and go "stop breaking our stats, please."

Some more quick-hitting thoughts on defensive play-action:

  • Coordinators matter. Wade Phillips left Denver, and they dropped from 14th to 31st in play-action defense. Wade Philips went to Los Angeles, and they rose from 26th to seventh. We've mentioned before how Wade Phillips brings immediate defensive turnarounds wherever he goes, and this is yet another example.
  • We talked about Carolina's improvement above, and New Orleans' improvement came with a great rookie class, but Detroit had the third-largest improvement without much in the way of new quality coverage talent. Rookie Jalen Reeves-Maybin was a bright spot, but as a whole, their linebacker group is not stellar in coverage. Instead, their improvement came on big splash plays. They led the league with seven interceptions on play-action passes, and added an above-average eight sacks, as well.
  • While there was a definite trend, not all of the top play-action defenses worked by forcing their opponents into checkdowns. Jacksonville and Washington ranked fourth and fifth in play-action DVOA, but were in the bottom ten when it came to aDOT. They found success in other areas -- Jacksonville by allowing just a 56 percent completion rate, Washington by leading the league with a 50.4 percent pressure rate.
  • You'd better be good at stopping play-action if you're in the NFC West; the Seahawks, Cardinals, and 49ers all ranked in the top eight in play-action frequency, as they had to deal with Sean McVay, Kyle Shanahan, and Darrell Bevell dialing up plenty of play-action. Of the three, only San Francisco got better when dealing with play-action. That's probably one part Reuben Foster and one part San Francisco's cornerbacks being terrible in general, so any trickery would have limited returns.

Comments

3 comments, Last at 20 Jul 2018, 9:35am

1 Re: 2017 Play-Action Defense

by ChrisLong // Jul 19, 2018 - 5:50pm

Has anyone ever looked at what exactly Wade Phillips defenses do better?

Login or Register to post comments

2 Re: 2017 Play-Action Defense

by Bright Blue Shorts // Jul 20, 2018 - 9:00am

Not much out there but first couple of pages of Google returns ...

this from his Texans days in 2011 ... https://www.battleredblog.com/2011/1/4/1913068/thats-so-crazy-it-just-mi...

And this from Denver ... https://www.milehighreport.com/2015/1/27/7922019/the-wade-phillips-3-4

The latter has the key insight that he fits the scheme to the talents of the players available. True of all the best coaches.

Login or Register to post comments

3 Re: 2017 Play-Action Defense

by Aaron Brooks Good Twin // Jul 20, 2018 - 9:35am

It feels like there needs to be some discussion of how Carolina was bizarro Baltimore.

Also, that MIN and JAX were good against both and that Indy and GB were terrible against both. With four teams, it didn't matter if you PA'd or not -- the outcome was similar either way.

Login or Register to post comments