Week 11, TNF Ravens-Bengals: Advanced Review

We'll find out the Bengals future with an MRI today, but they're in a tough spot to make the playoffs no matter the results

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

BAL vs CIN

The Ravens pull off a victory with the fundamentals closer than the actual score, and many long-term implications arose for both clubs. This game was one of the highest leverage contests of the week in terms of playoff probability, with Sumer Sports estimating before the game that the loss drove the Bengals playoff chances down to 36%, whereas a win would have put them at 59% (without factoring in injuries). The They estimated a win would raise the Ravens playoff probability to rise to 93%.

Both of those post-game estimate likely move down due to the injury implications of the game, although the Bengals’ playoff chances are hurt much more by any absence for Joe Burrow than the Ravens with Mark Andrews, whose season-ending injury is more of a factor capping the Ravens ability to hit a high-end outcome in the playoffs.

We should know later today the severity of Burrow’s injury, but even a quick return leaves the Bengals behind the odds. Looking across the larger NFL landscape, an AFC playoff bracket without Burrow, Josh Allen and Justin Herbert is a distinct possibility There’s a 35-40% chance they all miss the playoffs by my current numbers, with an assumption Burrow misses 1-2 games. This would be a PR and disaster for the league, with three of the games top-5 quarterbacks (by my numbers), all 27 years old or younger absent from the postseason.

Back to the game itself, the adjusted score differential of only four points reflects the fact that Bengals backup quarterback Jake Browning, a 2019 UDFA who had a single career pass attempt going into the game, performed unexpectedly well. Browning wasn’t good throwing the ball, but mixed in rushing value to help move the chains. The Bengals good play-by-play success was mitigated by third down struggles, going 3-for-11, including 0-for-6 after Burrow left the game.

The biggest play of the game in win probability and expected points added was the fluky tipped-pass-turned-touchdown for Nelson Agholor, though the Ravens were unlucky to have a 68-yard touchdown catch-and-run by Zay Flowers called back earlier in the drive by a questionable holding call.

The most consequential coaching decision was Zac Taylor’s call to put on 4th & 3 at the Ravens 48 yard-line with roughly 1:40 left in the second quarter. That call cost the Bengals 4.3% win probability, admittedly without an adjustment for the quarterback change. The combination of a reflexive instinct towards conservatism and an outsized focus on conversion rates made outside observers and the game’s announcers call going for a particularly risky call. This is a good decision to dive into because it requires thinking beyond “how do I feel about the chance of conversion”, which is the extent of a lot of analysis.

First, you have to view these decisions as exercises in shifting scenario analysis, not just conversion rates. The latter is the focus as it’s the most intuitive to think about, but the power of modeling comes from the ability to aggregate and process the numerous permutations for how the game could play out under differential decisions, and the associated post-play win probabilities if you kick, convert, or fail on the conversion. If you view a 4th & 3 decision as “do I feel good about Jake Browning converting this in a close game” (-4 point differential at that point), then you say no. But there are a multiple of reasons by scenario analysis that shift that reasoning.

I can’t go through everything that being processed in the model, but we can discuss a simple scenario analysis way of looking at things. The ultimate currency in the NFL is possessions. Each one gives you can opportunity to score points, the same for your opponent. In this case with 1:40 left in the half, you’re very close to the one time in the game when possession is reset, instead of being passed back and forth. End-of-half scenarios give teams a chance to gain in possession differential, which means gain in score differential without having to be any more efficiency on a per-possession basis.

If the Bengals punt with around 100 seconds to go, there’s a big chance they never get the ball back, the reverse if they go for it, i.e. they never give the ball back to the Ravens. So a conversion gains the Bengals a possession in great field position - just like any other point in the game - but also makes it likely to rob the Ravens of a pre-halftime possession. The Bengals were receiving the ball in the second half, so you could estimate that they’d have roughly a five-to-four-and-a-half possession advantage after halftime. Punting means a strong likely that shifts to 5-to-5.5 for the rest of the game, whereas a conversion would be more like 6-to-4.5 the rest of the game. That’s a massive shift. In the simplified scenario where each team averages four points per possession the rest of the game, the go-for-and-covert scenario means the Bengals win by 2 points, the punt scenario means the Ravens win by 6 points.

I get that Browning isn’t ideal for converting that one down, but I think we knew at that point there was a good chance he was going to be the quarterback for the rest of the game. Not trusting him to be better than the worst quarterback in the NFL on one play (basically the assumption you’d have to make, or harsher, to think punting is optimal), also means you’re putting him and the Bengals offense in situation where he’ll have to be relatively more efficiency that Lamar Jackson and the Ravens the rest of the game. Your win probability is very low with punting if you think Browning is that bad, so you’d want to increase the variance of win probability of a few plays and hope to get lucky. It’s more likely Browning will play well (which you need either way) on a few plays than dozens.

The Bengals probably lose the game no matter the decision, but a simple substitution of a pre-halftime touchdown for the Bengals after converting, and removing the Ravens pre-halftime touchdown closes the 14-point scoring gap. It would have been a lucky outcome for everything to work and win, but giving up the chance of hitting on that outcome certainly isn’t worth pushing your opponent back by by 27 yards of field position.

This was a strong efficiency game for Jackson, bouncing back after a mediocre game last week. Jackson continues his nearly every-other-week alternations of elite and meh numbers, but his accuracy and passing EPA have currently risen consistently this season versus prior years. An apparent ankle injury didn’t prevent Jackson from continuing to play.