2016 Play-Action Defense

by Bryan Knowles

On Wednesday, we looked at play-action offense. Now it's time to flip the script and look at the defensive numbers.

Every year when we write this article, people ask us to check to see if run defense is correlated with play-action pass defense. Some people think there will be a strong negative correlation -- that is to say, teams that sell out to stop the run will bite on play-action fakes more frequently. Others suggest that play-action is likely to be most successful in "obvious" rushing situations, so teams that play the run aggressively should get burned more frequently on the play fake. It seems to make logical sense, but does that theory actually hold water?

Well, not so much. The correlation between play-action DVOA and run defense DVOA in 2016 was .01. So much for that theory…

But that's just 2016. Here is the year-by-year correlation between play-action defense DVOA and run defense DVOA:

In odd years, the correlation between a team's run defense and their success against play-action is 0.41. In even years, the correlation is just 0.15. Clearly, residual fumes from the biannually lit Olympic torches reduces the ability of good defenses to accurately read play fakes, and 2017 will be a great year for run-defense play-action correlation…

OK, this is clearly just a weird random pattern; it doesn't hold up before 2009 and there's no logical reason why it would flip on and off randomly. It's likely a combination of A) random fluctuations (in the classic "correlation does not mean causation" sense) and B) the fact that defenses that are good at one thing are quite frequently good at other things, too. That's why you see a small positive correlation, not a negative one.

There's also no correlation between a team's run defense and their difference between defending play-action passes and non-play-action passes. You might think that teams that do well against the run play it more aggressively than other teams, and thus are more vulnerable to fakes, but there's no correlation there. Again, we see the odd-year/even-year splits here -- there's a 0.16 correlation in odd-numbered years and a -0.10 correlation in even-numbered years. Sometimes, stats are wonky.

No, we're overthinking things here a bit: play-action works against teams that don't have the personnel to stop play-action. That generally means poor linebackers and safeties; players who get sucked up to the line of scrimmage on the fake and lose track of the routes over the middle of the field.

That's exactly what you'll see in the league's worst play-action defenses; a Who's-Who of the worst pass-coverage linebackers in football last year. Detroit had Tahir Whitehead. Green Bay had Joe Thomas and Blake Martinez. New Orleans had Craig Robertson and Nathan Stupar. If I'm an offensive coordinator, those are the sorts of players I want to target with deception and misdirection, rather than players such as Benardrick McKinney and Brian Cushing in Houston, or Jordan Hicks and Nigel Bradham in Philadelphia. If you're looking for a correlation between something and strong play-action pass defense, look for the linebackers who can stay disciplined.

In the following table, 2016's defenses are sorted by ascending DVOA against play-action, with the best being at the top. Sandwiched around that is play-action rate and DVOA against non-play-action dropbacks. 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.

The league's overall play-action defense got a little bit worse in 2016, "down" to a 21.3% DVOA from 20.6% a year ago. (Remember, higher DVOA numbers mean worse defense.) It's the fourth straight year that stat has risen, and the league's worst performance since 2010 (22.2%). That maps, at least somewhat, onto the zone-read era we talked about in the offensive article, but at first glance, that seems more coincidental than anything substantial.

The correlation between play-action DVOA and non-play-action DVOA remains relatively low -- just .30 this year, down from .38 the year before. While at the end of the day covering a receiver is covering a receiver, there appears to be enough of a difference in recognizing and not biting on play-action deception to make a big difference in places like Houston and Green Bay.

Houston led the league with a play-action DVOA of -31.5%, which is impressive for two reasons. First, it's the fifth-best DVOA we've recorded this decade, behind only the 2015 Broncos, 2013 Seahawks and 2010 Packers and Jets. Those are some terrific defenses to be compared to, especially for a defense that only had a -6.9% DVOA overall last season. Second, Houston has been terrible at defending against play-action in recent history. They had a DVOA of 54.9% against play-action, third-worst in the league, in 2015. That wasn't a one year fluke either -- Houston was dead last in 2013, and hadn't put up a negative DVOA against play-action since 2011. Credit's due a lot of places here, I think -- Cushing returning somewhat to his 2011 form after a couple seasons of injuries and McKinney taking a huge stride in his sophomore season being two of the larger ones. It will be interesting to see how the defense continues to develop with the return of J.J. Watt. Houston's defense in recent years has been built around setting Watt up to be successful. The trick for 2017 will be re-integrating him without disrupting the improvements everyone else made last year.

Detroit finished dead last in play-action pass defense, but they also finished last in non-play-action pass defense, so perhaps there's no surprise there. Green Bay, on the other hand, was an adequate standard pass defense, but completely lost their way when play-action entered the equation. Their difference of 53.7% was dramatic, but not unprecedented; Green Bay had a difference of 40.1% in 2015 as well. Despite their back-to-back struggles, they only ended up facing play-action on 17 percent of their opponent's pass attempts. In fact, they were the only team in the bottom ten to face a below-average amount of play-action attempts. It's not due to their schedule, either; everyone else in the NFC North saw more play-action than they did.

You'd expect opposing offenses to attack that weakness more, but the Packers weren't alone there. The Broncos had the third-biggest difference between their play-action and non-PA DVOAs, yet they actually faced the fewest play-action pass attempts in the NFL. Meanwhile, San Francisco faced the most attempts, and yet defending play-action was one of the few things they weren't terrible at last season. This suggests that choosing to use play-action is more about accentuating your offense's skills rather than attacking your defense's weaknesses, per se.

Denver may have been coasting at least a bit on reputation -- they had the best play-action pass defense in the league in 2015, so opposing offenses may have been afraid to challenge them. They had the second-biggest dropoff in the league from 2015 to 2016. The worst dropoff, though, happened in Oakland, as the Raiders tumbled from -10.7% to 48.7%, or third-best to fifth-worst. This puts them more in line where they've been in recent seasons; 2015 was something of an outlier. No team gave up more yards on play-action than the Raiders did (9.7 yards per attempt). You can blame a lack of defensive discipline for that, as well as perhaps an overly aggressive pass rush from Bruce Irvin and the otherwise excellent Khalil Mack.

All this being said, it's important to remember that sustaining defensive success (or lack thereof) is difficult in the NFL, and wild swings from one year to the next are not uncommon. The year-to-year correlation on play-action defensive DVOA is just 0.07, so fans in Oakland or Detroit or Green Bay have every reason to hope their guys will bounce back in 2017. After all, it is an odd-numbered year.