2014 Pressure Plays: Defenses

2014 Pressure Plays: Defenses
2014 Pressure Plays: Defenses
Photo: USA Today Sports Images

by Sterling Xie

After checking out our charting numbers on how quarterbacks handled pressure last week, we're turning our attention to the other side of the ball and looking at defensive pressure DVOA. League-wide averages didn't change much -- pressure rate went from 24.8 percent in 2013 to 24.6 percent last season, while pressure DVOA (-76.0%) was essentially the same as last year (-75.0%).

On a larger scale, NFL defenses are creating more pressure, but with less success. The last two seasons have generated the highest pressure rates since 2010, but the defensive DVOA figures have been a little bit worse than the three-year average (-82.2%) from 2010-12. One might be tempted to pin causation on the officials' increased emphasis on coverage penalties, but DVOA only includes defensive pass interference, not illegal contact or defensive holding. And in the last two years, the average number of total DPIs (231) has only seen a slight uptick from the 2010-12 average (223), a minimal number when considering that those eight extra penalties are spread over 267 regular-season and postseason games. While the lesser DVOA is still nothing to sneeze at, pressure is another area where offenses are gradually making inroads and tilting the playing field in their favor.

Because of this, it's less surprising to see that creating pressure doesn't necessarily lead to good pressure defense. In 2014, the correlation between pressure rate and pressure DVOA was 0.06, while the pressure rate-non-pressure DVOA correlation was -0.02. In general, though, pressure rate hasn't shown much year-to-year consistency in reflecting how defenses will perform:

Pressure Rate Correlation w/DVOA
Year vs. Pressure DVOA vs. Non-Pressure DVOA
2014 0.06 -0.02
2013 -0.22 -0.44
2012 -0.04 -0.19
2011 0.32 -0.10
2010 0.17 0.16
Avg. 0.06 -0.12


14 comments, Last at 21 Jul 2015, 1:07pm

#1 by OSS117 // Jul 02, 2015 - 10:18am

Interesting read. So sacks make up about 25% of the pressures. Meaning hurries and forced scrambles make up the remaining 75%, obviously. And I would guess hurries make up the lion's share of that balance. Past several years Smallball has taken root and taken over. All the 'Check with me' plays, the package plays, the read/option pony sets, the hot reads, the WR/TE screens, the 3 and 5 step drops. The pocket/QB is getting closer and closer to the LOS. And by design, the ball is coming out almost immediately. There are no progressions. Most of these plays are based on the QBs presnap reads. He decides where to go with the ball based on the D before the snap, or reading the safety on his short drop. If the QB is fooled or makes an incorrect presnap read, he has to find another target quickly or bail. The protection isn't built to last. But I would suspect those plays where the QB has to find another outlet or scramble would be mostly unsuccessful. No disrespect to the charters, but it makes me wonder if hurries are being properly identified. On most of these aforementioned plays, it would appear that the QB is getting rid of the ball because of pressure. But by the nature of the play, pressure will almost always be present. But the QB isn't really getting the ball out because of that pressure. It's by design and by presnap decisions. Shrinking pocket depth and short setting protection, protection isn't expected to hold up for much more than the time it takes the QB to get to his short drop. And these are generally higher percentage plays. So it would give the appearance of both pressure and defeated pressure. Just a theory.

Points: 0

#2 by Joseph // Jul 04, 2015 - 9:14am

I think you're misunderstanding the stat here. The defense is getting pressure on about 25% of the pass plays--the other 75% of the time, the QB is not being pressured. This lines up with the upper 3rd of your post--plays are designed to get the ball out quickly.
Looking at the offensive stats (in the first part of this series), the top 10 LEAST-pressured QBs are: Peyton, Big Ben, K. Cousins (much smaller sample), Kyle Orton, Dalton, Rodgers, Stafford, Brees, Romo, and Rivers. (FYI, Brady is 13th) I think that we have a pre-snap reads group (PM, Ben, Brees, Rodgers, Romo, Rivers) and a scheme group (Orton, Cousins). Dalton and Stafford seem to be a combo of both and a good OL. We've all seen how PM covered up a horrible IND line his last couple of years there w/ great presnap reads.

Points: 0

#3 by OSS117 // Jul 04, 2015 - 11:38am

Sorry for the confusion. Filling in the blanks, if there was pressure on 24.6% of the 19942 plays, there was pressure on ~4906 plays. There were 1212 league sacks last year. 1212/4906 is ~25%. If sacks make up ~25% of that pressure figure, then hurries + forced scrambles make up the remaining 75%. There were ~1200 QB runs last year, not all of which were scrambles. But allowing that they were, they would make up less than 25% of those pressures. So 'hurries' make up over 50% of those pressure stats. Is that correct?

Points: 0

#4 by wmarsh73 // Jul 05, 2015 - 7:57am

There is an interesting note based on your statement

"Recall that a negative DVOA means good defense."

Looking at the non-pressure DVOAs - ZERO teams had a negative non-pressure DVOA while ALL defenses had negative pressure DVOA. If the above statement is accurate, it implies that the only good defense is pressure defense or am I misunderstanding something about defensive DVOA?

Points: 0

#5 by Vincent Verhei // Jul 05, 2015 - 9:32pm

Yes. If the defense gets pressure on the quarterback, it will usually turn out very good for the defense. If the defense does not get pressure on the quarterback, it will usually turn out very bad for the defense.

Points: 0

#6 by Thomas_beardown // Jul 06, 2015 - 10:49am

A defense that somehow gets the opposing offense to run instead of pass.

Points: 0

#7 by Paul R // Jul 08, 2015 - 4:12am

Re: Jacksonville and Tampa having so much success when they got pressure.

Watching the one or two Jacksonville games that were broadcast last season, I noticed that on pressure plays, an opposing quarterback would be so surprised to see a Jaguar player in his backfield that he would often drop his phone.

Points: 0

#9 by OSS117 // Jul 19, 2015 - 10:10am

I don't know if this will be seen/read since this thread has been dormant for a while, but I'll take the chance of asking anyways: Do you have the INT figures for these pressures, and absent the pressure? Because it looks like one of the most used/accepted axioms in all of football (INTs and Pressure go hand-in-hand) isn't true. There never seemed to be much if any relationship before between sacks and INTs. Tho obviously the QB isn't throwing INTs on sacks. But that's all we had to go on as a reflection of pressure. Until now. However, the pressure data shows the weakest of correlations to INTs at less than .2. It would be helpful if the scrambles could be filtered out, and sacks too, leaving just hurries. So how many INTs were thrown on hurried pass attempts, and how many on unhurried pass attempts. I've never purchased the almanac before. If all this is in there, just say that and I'll be sure to buy it.

Also interesting in the above data, Jacksonville was the worst in the league at generating pressure but was tied for 5th in sacks. Didn't generate many chances to get to the QB, but when they did they sealed the deal. The anti-Brett Keisel, who his coaches lauded every year as the king of their mystery in-house 'pressure' stats, crediting him annually with 35-45 pressures, while only managing a couple of sacks. I guess that award would go to the Chargers, who finished 11th in pressures, but ranked 29th in sacks with 26. Or Oak, who although ranked higher in pressures, had an even lower percentage of sacks/pressure.

Anyways, interesting stuff. As mentioned, I'm particularly interested in the relationship between pass attempts under duress and INTs. So any clarity on that would be appreciated.

Points: 0

#10 by Thomas_beardown // Jul 19, 2015 - 11:49am

I think INTs are a rare enough event you have trouble finding any correlation to them.

As for Jacksonville, I wonder if they were so bad QBs were told not to make any risky throws no matter what. Just eat the sacks, make sure we don't turn it over and know their offense won't do anything.

Points: 0

#11 by Vincent Verhei // Jul 19, 2015 - 11:39pm

For the sake of simplicity, let's just look at receptions, incompletions, and interceptions, and throw out penalties and laterals and intentional grounding and other weirdness.

Passes in 2014, no pressure: 311 INTs in 14,325 passes (2.2% INT rate)

Passes in 2014, pressure: 137 INTs in 3,456 passes (4.0% INT rate)

So the rate of INTs nearly doubles under pressure, which is a lot.

There is data like this in the Almanac. I can't promise anything this specific, because if we included every single statistical nugget the book would just be a 1,500-page spreadsheet with no actual writing. But we have a lot of stuff like this in there.

If all you want is mass data, you could also purchase the charting data, which is an Excel file, and study everything in there to your heart's content. Data for 2014 is not on sale yet and probably won't be until after we finish the book. But 2011, 2012, and 2013 can be had right now.


Points: 0

#12 by OSS117 // Jul 21, 2015 - 8:40am

"I think INTs are a rare enough event you have trouble finding any correlation to them."

IMO, it probably has more to do with it being a near-random independent event. A coin doesn't care what time of day or surface it lands on.

"Passes in 2014, no pressure: 311 INTs in 14,325 passes (2.2% INT rate)

Passes in 2014, pressure: 137 INTs in 3,456 passes (4.0% INT rate)"

Thanks for that. That is interesting. If I'm reading this correctly, you are using strictly every forward pass attempt on a credited play. So you have filtered out sacks and scrambles. And you're right, that's not peanuts.

It is interesting that in this article it's suggested that pressure is trending up. And the last two year were high-water marks. Pretty much everything associated with the passing game is trending up the past decade and a half. Attempts, completions, completion percentage, yards, ypa, etc. Everything, that is, except INTs. Those are going the opposite direction. Hard.

Aside from the strike shortened '82 season, last year the league totaled the fewest INTs since merging into a 26 team, 14 game league in 1970 (nearly half as many league pass attempts then). Turn of the millennium, INTs were 1 out of every 30 pass attempts. Last year it was 1 in 40. Pretty huge swing in a relatively short period of time. And that nosedive took a steeper angle about the time Goodell took his throne at 345 Park Ave.

With INTs trending down, and if pressure is trending up, then obviously pressure isn't the cause of this decline in INTs. There was a -0.2% diff in pressure from 2013 to 2014. While INTs went from 502 to 450. If that same pressure was maintained last year, INTs extrapolate to 451.5. The diff in pressure doesn't come close to accounting for the diff in INT.

Only theory I have for this is the smallball league this has become, and the change in the way the game is officiated/garnished now.

Kinda surprise tho that this trend is getting no notice/attention anywhere. Especially on a stat gristmill such as FO. IMO, this is a pretty big deal/story.

Fwiw, the only thing that I've found that correlates to INTs is subpackage usage. I know subpacks are mostly used in passing situations, so INTs will naturally occur more frequently there. But last year 71% of all pass attempts were against subpacks, while 79% of all INTs were made in subpacks. I don't have accurate data from previous years, so maybe that overage can be explained by a hotstreak in that snapshot/chain. But 450 INTs should be enough to smooth that out. So it would seem that subpacks do make a difference. But, perhaps like pressures, not enough to offset whatever is happening that's causing INTs to bleed off.

Points: 0

#13 by jtr // Jul 21, 2015 - 11:17am

I think the extra INTs in the sub package reflect a bias in personnel usage based on situation. The offense is more likely to be in their "base" offense in situations they're not going to take much of a risk: 1&10, 3&1, goal line, nursing a lead, etc. While they're more likely to be in 3+WR in situations they're more likely to try a high risk/high reward play: 3&long, 2-minute drill, trailing by multiple scores. Plus, it just makes sense intuitively that INTs become more likely when the defense substitutes to put an extra pass defense specialist on the field.

Points: 0

#14 by Thomas_beardown // Jul 21, 2015 - 1:07pm

It's clearly established that the talent of the QB and his tendencies is the largest indicator of INT%. So yes, pressure is not the biggest contributor, that doesn't mean it's not a contributor at all.

Points: 0

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