2014 Play-Action Defense

2014 Play-Action Defense
2014 Play-Action Defense
Photo: USA Today Sports Images

by Sterling Xie

Last week, we looked at offensive play-action numbers from 2014. Now we'll flip to the other side of the ball and look at defensive play-action numbers. Again, we would like to thank ESPN Stats & Information, from which we get our data on play-action.

Prior renditions of this column have often come with lots of shoulder shrugs. We know by now that there's very little year-to-year correlation regarding how defenses perform against play-action. This year's league-average DVOA against play-action was 15.9% -- slightly higher than the 12% to 14% range we have found in the past, but not enough to raise eyebrows. Defenses continued to fare worse against play-action (the league-average DVOA against non-play-action passes was 5.4%), but 13 teams fared better against play-action, suggesting that a mass increase in play fakes wouldn't really ambush defenses.

This also highlights the surprisingly low correlation between play-action DVOA and non-play-action DVOA, something Scott Kacsmar discovered last year. The 2013 correlation of 0.38 was virtually unchanged this year, falling slightly to 0.34. Even though a relationship still appears to exist, it's hard to reconcile why the Seahawks pass defense that was so great against play-action in 2013 would be just average one year later, while the dreadful Jets and Bucs units each transform into a competent above-average defense against play-action.

There's no "magic bullet" formula that encapsulates how a defense will fare against play-action, but we can look at particular characteristics and see if they provide any hints. One idea here could be to search for a connection between play-action defense and teams that have great pass rushes. Last year's article illustrated how the average depth of play-action passes was significantly longer than non-play-action passes, a finding that's hardly surprising. With slower developing routes featuring deep dropbacks, strong pass-rushing defenses should theoretically have a better shot at disrupting these big-play opportunities. So going back to 2007, I took a look at the correlation between adjusted sack rate and both play-action DVOA and percentage of play-action passes faced:

Correlations with Adjusted Sack Rate, 2007-14
2014 -0.24 -0.18
2013 -0.22 -0.30
2012 -0.20 -0.18
2011 -0.02 -0.33
2010 -0.45 -0.21
2009 -0.27 -0.36
2008 -0.24 -0.47
2007 -0.23 -0.02
Avg. -0.23 -0.26


16 comments, Last at 28 Jun 2015, 4:07pm

#1 by theslothook // Jun 23, 2015 - 2:05pm

is the dvoa calculation for PA done after the PA play has been executed? Ie - take the result of the PA play and then put it through the dvoa matchinations, or does it adjust the PA before by taking into account the down, distance, and game conditions?

Points: 0

#2 by Vincent Verhei // Jun 23, 2015 - 2:15pm

I don't understand your question. We have DVOA calculated for every play. DVOA always takes into account down, distance, and game conditions. Play-action DVOA is just each team's DVOA on plays that were play-action passes.

Points: 0

#3 by Aaron Schatz // Jun 23, 2015 - 2:59pm

I think I might understand your question here. Are you asking if the baseline that represents "average" incorporates the fact that this is a play-action pass along with all the other things that it accounts for? The answer is no, which is why the league-wide DVOA for play-action passes is above 0%. The DVOA for play-action passes only incorporates the things usually incorporated into DVOA (i.e. down, distance, and game conditions). And with team DVOA usually, "average" accounts for both passing and rushing, so in general teams will be positive passing/vs. the pass and negative rushing/vs. the run.

Points: 0

#5 by MC2 // Jun 23, 2015 - 6:56pm

Have you compared PA DVOA for defenses to run defense DVOA? Intuitively, you would expect to see a negative correlation, as teams that sell out to stuff the run would seem to be more vulnerable to PA, although a cursory glance at the table doesn't really seem to bear that out. It looks like, of the top 8 run defenses from last year, 3 actually did better against PA, and a couple were about the same.

Interestingly, the only 3 teams that were really good against the run and really bad against PA were all from the NFC West (Arizona, Seattle and St. Louis).

Points: 0

#6 by jtr // Jun 24, 2015 - 8:51am

I was wondering the same thing. Just anecdotally, I'm not surprised to see the Steelers did poorly against PA since they weren't doing well against the run. My enduring mental image of last year's Pittsburgh D was of the Browns gashing them with zone stretch, leading to the Steelers stacking the box and getting beat by PA.

Points: 0

#7 by OSS117 // Jun 24, 2015 - 9:24am

That was exactly the first thing I thought of, too.

Steelers were way too concerned about stopping the run last year, when they should've been more concerned with how crappy their pass defense was in their overused base 34.

The next thing that jumped out was the Jets and Eagles. Both were also overly committed to stopping the run last year. Jets led the league by a wide margin in Base D percentage at 62.2%. Eagles were next at 56.4%. You would think they would be ripe for play-action. But they were both close to the bottom in percentage of play-action plays faced. Seems nobody really tested them. Tho the pass defense in Base v Subpacks shows little to no difference, so maybe it just looks like an overcommitment to stopping the run, when in reality the 5th and 6th DB offered no benefit so they chose to just stay in base more. Dunno. Both were pretty good v play-action and kinda crappy v no play-action. I guess nobody felt the need for it.

Not the case with the Steelers. They were in Base 51.1% despite little difference in Run D (base v packs), while their Base Pass D surrendered 8.22 ypa, compared to the 6.50ypa while in nickel. I don't get their obsession with the run, and secondary concern with their secondary.

Points: 0

#8 by jtr // Jun 24, 2015 - 11:18am

The idea that the Steelers have to make stopping the run their #1 priority was a mainstay of Lebeau's philosophy, and probably part of why some of his schemes looked stale the last two years. I don't have much confidence that will change, as Butler was a Lebeau protégé who seems unlikely to mess too much with their existing schemes.

I'm looking forward to seeing the offensive formations faced by each defense in the upcoming FO Almanac. It doesn't really make sense to analyze defensive personnel usage without that data, since base-vs-nickel decisions are usually determined entirely by offensive personnel. I have a feeling the Steelers 49% sub package usage corresponds to facing offenses in 3+ WR sets almost exactly 49% of the time. That extra corner doesn't do you any good if he's going to cover a fullback who isn't running a route anyways.

Points: 0

#11 by runaway robot // Jun 24, 2015 - 2:15pm

Don't forget that Tomlin made his bones on defense. I don't think Tomlin, when he took the Steelers job, expected the then 70-year-old LeBeau to be his DC for eight more years. I also think Butler, at this late point, has to be anxious to make a name for himself. Maybe I'm just looking for reasons to be hopeful, but I do believe the Steelers are about to make a real change in defensive emphasis.

Points: 0

#12 by jtr // Jun 24, 2015 - 3:16pm

The fact that they've drafted some damn defensive backs is certainly a reason for optimism. But I don't think promoting a linebacker coach (and former pro LB) to defensive coordinator leads to LESS of a misguided emphasis on defending the run. Ultimately, I think the problems under LeBeau were more about lacking talent and speed than they were about scheme, and the team has not been shy about investing their draft picks on defense lately.

Points: 0

#16 by OSS117 // Jun 28, 2015 - 4:07pm

It's kind of a chicken/egg dilemma, tho. Did they play that much base because opponents played that much 2TE? Or did teams play that much 2TE against them because they knew LeBeau would respond with that much base?

You're probably not too far off specifically regarding the Steelers 2TE and 3 WR base/nickel matchup percentages. But other things factor, or should factor, into those match up decisions. Score, time, down/distance, and tendency. It remains tho that the league passes ~60% of the time and runs ~40%. Yet the Steelers were rolling out a run defense over 50% of the time. 1st and 10 is about a coin flip run or pass. 2nd and 5 is about 50/50. Above that it's about 2 passes for every 1 run. 3rd and 2 is the tipping point for run/pass. Those down/distance markers are better indicators for run/pass than formation anymore. Or, better yet, combined to better dial in the proper D.

I suspect teams used more run formations against the Steelers because they knew LeBeau would match it with Base most of the time, giving the O favorable pass match ups. And because their Run D also stunk. Best of both worlds. If I had to guess I'd say opponents had a higher usage of 12, 21, and 13 against the Steelers than they did in the rest of their games.

TEs can't really block well anymore, even the #2 TEs. They still create extra gaps to defend, but I don't think teams should get too hung up on rigid match ups. Especially if the TEs are more like WRs anyways. If you know the other team is going to pass 60/40, you're nickel/base decisions ought reflect that. Or try. The Steelers base defense gave up 8.22 ypa against the pass. Compared to their nickel 6.50. And their stats against TEs were pretty bad. LeBeau was too hung up on stopping the run, and married to Base v 2TE. Yeah their run D was bad, but their pass D in base was much, much worse.

I don't have those opponent percentages. But, incidentally, you can get pretty close with the snaps listed in gamebooks. You know how many plays they ran. Just count the total snaps by unit (TE, RB, WR). If a team plays 60 offensive snaps, and their TEs total 75 snaps, they were in 2TE ~15 snaps. Same idea with RBs. Or if the WRs total 150, they were in 3WR ~30, etc. 4 and 5 WR, and 3TE or extra OT, etc will skew it a bit. But predominantly it's 12, 21, or 11, so it gets you pretty close. Those overages should come really close to reconciling with total snaps.

Same on the D side, count the total DB snaps. If a team played 60 D snaps, and the DBs totaled 280, they were in nickel ~40 snaps. Dime, jumbo, etc will skew it, but again, should get pretty close. Some teams that play more dime than nickel will screw it, and teams like Arizona who play a lot of 7DB packs really screw it. But most of the league plays predominantly nickel subpacks. So, close enough for a quick ballpark gauge.

It's not too hard to get a team's own personnel groupings exact. On either side of the ball. But outside of charting, opponents are much tougher to mine. So the gamebook trick is better than nothing. Better to spend 5 minutes getting to the ballpark than 500 hours getting to the batter's box. Or zero minutes and remain in the dark.

Cleveland used a ton of 12, 21, and 13 against the Steelers last year. And almost no 11. And the Steelers responded by playing very little nickel. On the season Cleveland did combined for a lot of 12, 21, and 13. About 550 of their 1011 plays. So about a 54/46 split in run formations v pass formations (3WRs or greater). But if you subtract out both Steelers games it changes to about 47/53, roughly a 130 play swing. Pretty significant jump against the Steelers in run formation usage.

Their nickel run D got all the bad local press last season. But it was their base 34 that their opponents tried to keep on the field. After the dredful preseason, understandable LeBeau's lack of faith in the nickel run D. But after their base pass D kept getting habitually flogged, I have no idea why LeBeau stayed committed to keeping his base 34 on the field as much as he did. Their nickel run D righted the ship early in the season (almost level with their base run D). That base pass D never ever did.

Points: 0

#9 by mehllageman56 // Jun 24, 2015 - 12:11pm

I don't know about the Eagles, but the Jets secondary was so terrible that there may have been times playing a linebacker at one corner would have been a good idea. The safeties weren't that bad, but they really had no corners. The other thing to think about is that play action would allow more time for the linemen to pressure or sack the qb, which is the only way to stop a passing attack when your corners are so bad.

Points: 0

#10 by theslothook // Jun 24, 2015 - 1:21pm

Im not really convinced we can parse which groups PA really targets. Can you make your PA defense work if your safeties stink but your corners are solid/linebackers/dl line? We don't even know how to parse credit/blame for non PA, garden variety pass defense, to say nothing of how pressure interacts with those groups.

Points: 0

#13 by mehllageman56 // Jun 25, 2015 - 9:12am

Going by the cornerback charting post earlier in the offseason, Jets opponents did well throwing against any of their cornerbacks. All play action did against the Jets was make the quarterback take more time to throw, thus enhancing the chances the Jets linemen got to disrupt the play. I might add, having watched many of the Jets games last year, teams were targeting whichever corner couldn't cover their man, which was often all of them.

Points: 0

#15 by theslothook // Jun 25, 2015 - 1:29pm

Yeah but doesn't arizona have good corners and they stunk against Pa?

Points: 0

#14 by jtr // Jun 25, 2015 - 12:06pm

I think the conventional wisdom is that PA targets the linebacker group in particular. LBs get sucked up to the line of scrimmage chasing the run fake and lose track of the routes in the middle of the field. So linebacker discipline is probably the #1 factor in PA defense. And looking at the top of the list, SF, KC, TB (at least David) and CAR are all teams with good off the ball LBs, though obviously SF was not at full strength at the position this last year.

Points: 0

Save 10%
& Support Sterling
Support Football Outsiders' independent media and Sterling Xie. Use promo code WRITERS to save 10% on any FO+ membership and give half the cost of your membership to tip Sterling.