Analysis: Is the Buffalo Bills’ defense hitting its stride?

Aaron Schatz of the much maligned (by Buffalo Bills fans) Football Outsiders tweeted an intriguing statistic suggesting a drastic turnaround for the Bills’ defense. In their proprietary metric of Defense-adjusted Value Over Average (DVOA) the Bills ranked 30th through their first six games. In the last five they’ve shot up all the way to seventh place. DVOA is an aggregated statistic that looks to measure success rates and factors in numerous factors such as down, distance, and quality of opponent. A full definition can be found here.

While Football Outsiders always deserves credit for their number crunching, we decided to take a look at some data ourselves to check the turnaround across a variety of metrics. For the sake of consistency we used the same split (first six games versus the last five).

---

The Sexy Stats

First let’s discuss the charts. I’ve tried to color code for quick comparison between the two groupings. Individual game results are included for review, but for the purposes of this dive it’s the average of the first six games up against the next five. Green cells indicate that the grouping has had better performance and red means worse performance.

This is pretty straightforward. In every one of these stats the Bills have done better in the most recent five games. In most cases there’s a pretty significant change.

• The Buffalo Bills have been much better at getting to the quarterback, more than doubling their sack rate. Sam Darnold and Russel Wilson likely left the game quite sore.
• Quarterback hits increased by nearly 50 percent, which is another huge jump. Again, poor Darnold and Wilson.
• Passes defended and tackles-for-loss both increased significantly and both have been semi-consistent between games.
• The Bills have been very steady in turnovers all year, but in the last five games have avoided any zero turnover games and forced a huge cluster against the Seattle Seahawks.
• Explosive plays, anything going for 20 or more yards, have remained mostly stable and represent the only sexy stat that didn’t have major improvement.

Yards

Overall yards

You’ll notice I added some gray cells here. They’ll be used in all the remaining charts. They represent the opponents’ average. Essentially they’re a way to adjust performance by quality of opponent. Using the raw numbers look at the games against the New England Patriots and Los Angeles Chargers. On the surface the Bills did better against the Patriots, allowing 18 yards less.

However the Chargers gain 50 more yards on average than New England does, suggesting that the Bills’ defense actually did a better job slowing down Los Angeles. In essence, it’s the “Difference” between what actually happened and what would be expected based on the opponent’s usual performance.

• In terms of absolute yardage per game, the Bills’ defense allowed about 32 yards less in the more recent contests—which is a decent shift.
• Looking at the gray cells, both groups of opponents averaged a very similar yardage total per game.
• In the first six games, the Bills averaged 31 yards worse than what would be expected based on opponent averages. The Miami Dolphins and Los Angeles Rams in particular gained a lot more yards against the Bills than they do on average.
• In the next five games the bird-themed teams gained more yards than their average but that’s about it. The Patriots broke even and the Bills held the Jets way under their usual and the Chargers a bit under their usual (and that’s even with a successful Hail Mary factored in).
• When adjusting for opponent, the Bills improved 36.8 yards per game. This also flipped them from the wrong side of “break even” to the right side, if only slightly.

Yards Per Drive

I like yards per game but I don’t love it. An effective offense can limit opponent opportunity and make the defense look better. Yards per drive isolates that factor away and only considers what happens on average every time the opponent has the ball. It’s strictly Bills’ defense vs. opponent offense for this.

• Using the raw numbers, the Bills improved by 6.6 yards per drive. That doesn’t sound like much, but 39.9 would currently rate worst in the league, while 33.3 would put the Bills at 14th.
• For the gray cells, the average team gains 32.9 yards per drive, meaning that both groupings are above league average and overall pretty similar.
• For the opponent-adjusted cells on the right-hand side we see the same pattern as we did for yards per game. The Bills have not only improved significantly but they’ve gone from allowing more yards per drive than should be expected to slightly less.

Yards Per Play

Drives do have the potential for statistical wackiness. The highest sample size for a game this year has been 12 drives. For plays the highest is 83. That means a weird item or two has the potential to throw off drive-based data faster than play-based data. (Something like a kneel down.) A two-play drive with two knees still counts as one drive (so 1/12 in our example). For plays it would only be 2/83 in our example. Yards per play helps eliminate that possible anomaly and we do get an interesting result.

• The Bills again do better in the absolute metric but, intriguingly, their opponent averages (gray cells) suggest that the second grouping was actually a tiny bit worse than the first six-game grouping. This is the first stat to suggest that situation.
• Just as odd, this is the first stat where the Bills’ defense didn’t give up more than expected yardage, breaking exactly even in the right-hand columns.
• Forgive the weird rounding on the -0.4 in the “Difference” column. The exact number is -0.38 as the Bills have allowed 5.16 yards per play in these games and the opponents’ average is 5.54 yards per play.
• The league average is 5.5 yards per play, meaning the Bills have overall faced an average set of opponents (the Jets are so bad they balance out multiple good teams on their own).
• I’d encourage everyone to look up yards-per-play stats beyond this. An improvement of 0.38 yards per play is pretty big.

Points Allowed

Of course points are what matter the most. Here we have another measurement where the second group of teams has performed worse than the first six-game grouping.

• The Bills have improved by five points per game in absolute terms.
• They’ve improved by 3.8 points per game when adjusted by opponent.
• Where they were allowing on average about one field goal a game more than you’d expect, they’re now limiting teams to just below their normal scoring output.

Points per drive

Just like with yardage, the offense has some ability to dictate how many chances an opponent gets. Points per drive helps isolate that. We won’t go all the way to the per-play level like we did with yards as this is getting long enough. I did run the numbers and the story we see keeps repeating anyway.

At this point it’s pretty clear. The Bills have improved in their absolute numbers pretty much across the board on defense. Adjusting by opponent doesn’t change this and adds an additional layer that we see here yet again. Where the “early 2020” Bills were allowing more yards/points than expected based on opponent averages, the “recent 2020” Bills have started to pull these numbers below the expected.

It’s way too early to crown them elite, but the turnaround is real.