Coverage stats show flaws in L.A. Rams defense

Over the past two weeks, we have looked at both YAC+ and our plus/minus stats from an offensive point of view, highlighting Deebo Samuel and Cooper Kupp and Kyler Murray and the best players in the passing game. But none of these stats are adjusted for the skill of the defense. That raises the question — should they be adjusted for quality of opponent? Is there such a thing as a YAC-preventing defense? Do better pass defenses produce lower CPOEs above and beyond their control over where the ball is thrown? And if there are, is that predictive from year to year, or is it dependent on the strength of the offenses they face?

My hypothesis coming in was that we wouldn’t see any real consistency in YAC+ data from the defensive point of view, that an offense’s ability to generate yards after the catch would solely be based on their own play designs and the skills of their players, and thus defense-adjusted YAC+ wouldn’t make a lot of sense. I believed we would see some consistency in plus/minus and CPOE, but not in a way distinct from general pass defense DVOA; a defense wouldn’t be particularly better or worse at preventing completions as they would be at just stopping pass efficiency in general.

But any good hypothesis needs a test, so let’s dive into 2021’s data and see what we can find.

These numbers are still from the offensive point of view, so negative numbers are better for defenses.

First step to see if this means anything: check the correlation with pass defense DVOA in general. Both plus/minus and CPOE have a 0.74 correlation with pass defense DVOA, and that correlation stays at roughly that level in prior years. Good pass defenses prevent opponents from completing passes; yet another breakthrough in analytical knowledge from your friends at Football Outsiders.

It is interesting to see where DVOA and CPOE diverge. There are three teams who dropped 10 or more ranks between DVOA and CPOE — in other words, they were significantly worse at stopping completions than in playing pass defense as a whole. Those were the Rams, Panthers, and 49ers. I think what we’re seeing there is a strong pass rush component to their pass defense — the Panthers were third in adjusted sack rate, the 49ers fourth, and the Rams eighth. This makes a lot sense for the 49ers and Panthers — San Francisco was forced to start Josh Norman for part of the season due to injuries and poor performance, while Jaycee Horn’s injury meant Carolina had no solid boundary corners to speak of for most of the season. It’s no surprise that the units led by Nick Bosa and Hassan Reddick would be the strong points of those particular defenses. The Rams are a little harder to explain, but even Jalen Ramsey only was 25th among qualified cornerbacks in success rate last season — you’ll find that stat and more in Football Outsiders Almanac 2022! — while neither Darious Williams nor David Long exactly covered themselves in glory. Aaron Donald and company were certainly better than the secondary, though I wouldn’t have expected the Rams to fall all the way to 27th in CPOE.

On the flip side, four teams jumped 10 or more ranks between DVOA and CPOE, and so were better at stopping completions than their overall pass defense would indicate. Those are the Ravens, Broncos, Packers and Falcons. The Falcons and Ravens were both in the bottom three in adjusted sack rate, so those drops make total sense — they had few sacks and forced throwaways, so it logically follows that the vast majority of their (limited) pass defense success would come from great plays from AJ Terrell or Marlon Humphrey. The Broncos and Packers were both below average in ASR, but not significantly so, so that analysis breaks down a little. For the Broncos, that might be the rookie performance by Patrick Surtain coupled with very solid coverage from non-corners such as A.J. Johnson and Justin Simmons. For the Packers, there may be a strength of schedule component here, as getting two games against Justin Fields, a game against Sean Mannion, and games against banged-up Russell Wilson and Baker Mayfield probably help.

YAC+, on the other hand, has just a 0.10 correlation with pass defense DVOA. We can try to squint to see if we see anything meaningful there. The Ravens were excessively injured, so perhaps it makes sense that their defenders were more likely to be out of position or bad tacklers and explain their league-worst +1.1 YAC+ allowed. Maybe you can argue that the Raiders were fast under Gus Bradley, and that helped them stop runners in their tracks by flowing to the ball better than most teams? They were ranked the 10th-fastest defense in the league from ESPN, which could help chase down receivers…

No, I think the simpler answer is that defensive YAC+ numbers don’t tell us anything meaningful. If the ability to stop YAC was a skill, we’d expect to see some year-to-year correlation in these numbers, but there just isn’t any. There was a -0.06 correlation in a defense’s YAC+ between 2020 and 2021, which is a rounding error away from zero. Other years see higher correlations, but not frequently enough to suggest it’s anything more than random chance, even taking into account the inherent variability which comes with just 32 datapoints and no control. Offensively, YAC+ consistently sits around a 0.36 year-to-year correlation, which is low but reasonable when you consider how much it depends on scheme, quarterbacks, and particular receivers. YAC is noisy, for certain, but what signal is there is almost entirely focused on the offensive side of the ball. This makes a fair degree of sense — offenses get to control their play calls and targets and can choose to draw up plays to get Deebo Samuel or Ja’Marr Chase the ball in space. Defenses just don’t have that kind of control over the type of plays or players they face.

The year-to-year correlation of defensive CPOE isn’t that hot, either — just 0.18 last year, about average for this stat. I think you can use it a little bit descriptively, to see which teams relied more heavily on their pass rush or coverage skills than others, but I don’t think there’s really enough there to justify adjusting CPOE based on defenses going forward. On the offensive side of the ball, there’s a 0.55 year-to-year correlation in CPOE, and that number gets higher if you only include teams that kept the same quarterback. The most significant factor in a pass being completed remains the accuracy of the passer.

In short, these defensive numbers look less significant than their offensive counterparts — which is why we’re putting them in one article here rather than breaking them down individually. While there are some interesting tidbits that can be taken from these numbers, both YAC+ and plus/minus appear to be primarily offensive statistics.