Week 10, MNF Bills-Broncos: Advanced Review

The Bills lose another game they should have won, and now playoffs aren't guaranteed

The adjusted scores quantify team play quality, with emphasis on stable metrics (success rate) and downplaying higher variance events (turnovers, special team, penalties, fumble luck, etc). Adjusted expected points added (EPA), in conjunction with opportunity-based metrics like total plays and drives, projects adjusted points. Adjusted scores have been tested against actual scores and offer slightly better predictive ability, though their primary benefit is explanatory.

All 2023 & 2022 and historical Adjusted Scores and other site metrics are available in a downloadable format to paid subscribers via Google Sheet.

Find previous advanced reviews here

** Adjusted Scores table:

• “Pass” - Pass rate over expectation (based on context of each play and historical averages

• “Success” - Success rate on offense, a key metric in adjusted score vs actual

• “H & A” - Home or away team

BUF vs DEN

Another week, another Bills loss where they were the superior team by the fundamentals-based adjusted scores. I decided to look at seasonal “records” based on adjusted scores (going down to a tenth of a point to reduce the incidence of ties), and, amazingly, the Bills would be 10-0, not 5-5, using these adjustments. Are the Bills fundamentally at the strength of a 10-0 team? No, that’s not what I’m saying. It’s a combination of a team that is much better than 5-5, plus enough close games to mean that a parallel universe of good luck for the Bills could push a number of wins into their column.

The inevitable issue presenting these numbers to the public is the response that there’s something about the Bills’ DNA that isn’t captured by the numbers. There’s something inextricably linked to the team that causes them to lose games like last night, when they have an offensive success rate vastly superior to their opponent (48% to 37%). It only takes a few results of this nature for many to come to this conclusion, though a broader view of years of evidence shows us that variance-dependent (i.e. luck-based) results can persist in one direction for much more than a few games, and even longer than an entire season.

One of my favorite examples of Bills-related media panic based on small-sample “truths” was found in the aftermath of their weather apocalypse loss to the Patriots Week 13, 2021. The Patriots came to Buffalo and won 14-10, while throwing only three passes in extreme wind. The Athletic’s postgame headline read: “In Patriots’ latest win over Bills, Bill Belichick shows he’s living in Sean McDermott’s head”. In the article, we were told that “what the Patriots did to the Bills psychologically won’t be easy for Buffalo to overcome”, and “the Bills don’t have the team to handle late-season games against physical opponents.”

I guess the Bills just folded the rest of the season then, right? Instead, following an overtime loss to Tom Brady in Tampa, they ran off five straight wins, and were arguably the better team in a crazy 42-36 overtime loss to the Chiefs in the Divisional Round. Those wins included two convincing victories over the Patriots, with a combined margin or victory of 42 points! Nothing we we told about the Bills based on a handful of mid-season results ended up being true, at all. Instead, they regressed to, and maybe even outshot in the other direction, their fundamentals as a great team. I expect something similar the rest of this season.

All that said, the Bills are 5-5 and have one of the most difficult schedules the rest of the season. Those results happened, justified or not. This matchup against the Broncos, at home, was supposed to one of their few easier remaining games. They have a gauntlet left on their calendar: NYJ, @ PHI, @ KC, DAL, @ LAC, NE and @ MIA. Right now, implied probabilities based on betting odds have the Bills’ making the playoffs roughly 30% of the time.

Now to the game last night. As I mentioned, the Bills were the superior offense on a play-by-play basis, though their passing attack had one of the worst games of the year, even by fundamentals. The Bills lost all three high-variance aspects of play that I highlight in the first table: turnovers (-12.7 EPA), late-down conversions (-12.7 EPA) and special teams/penalties (-3.9 EPA). We don’t throw those results completely out the window, but we downweight more random stuff to have better predictions into the future.

The Bills lost two of three fumbles, which were more costly situationally than the one Broncos lost fumble (1st & 10 vs 3rd & 6). The most impactful play of the game for expected points was the Bills loss of an apparent option exchange between Josh Allen and James Cook, with the former inexplicably losing control of the ball, an entirely avoidable 5.6 expected points loss.

The Broncos converted two more first downs than expected based on distance in the game (8-of-18), after being slightly below average the rest of the season. The Bills converted third down below expectation (3-of-8), after doing so well over expectation earlier this year and the last few seasons. Plus, the Broncos gained a net 7.5 EPA on the Bills by converting their one 4th & 2 try, while the Bills couldn’t covert their 4th & 1.

Russell Wilson’s numbers very much match what “good Russ” would do with the Seahawks: translate enormous good accuracy (+13.3% CPOE) into great-not-elite efficiency. Russ’ aDOT was an absurdly low 3.4 yards, yet the high completion rate and YAC production made it a good night. As he’s known to do, Russ still took four sacks that cost the Broncos 5.2 expected points, roughly equal to the losses on Allen’s two interceptions (-6.0 EPA), which will get much more negative scrutiny.

There’s no two ways about it: this was a bad game for Allen. You can’t point to this game to prove that previous EPA-based numbers that liked Allen this season are misleading, because those same EPA-based numbers hate this performance. The disappointing thing about this loss for the Bills is that the running game and defense came through well enough to provide the victory, despite Allen having an off night. This was, by far, the worst efficiency game of the season for Allen, only have slightly negative EPA (-0.02 per play, Week 1 vs NYJ) in his second-worst performance.