Week 14 TNF Review: Baker to the rescue

An ugly win for the new Rams quarterback, made possible by Josh McDaniels' conservatism

*The adjusted scores attempt to quantify team play quality, with emphasis on the most stable metrics (success rate), while downplaying the higher variance aspects of play (turnovers, big plays, special team, penalties, fumble luck, etc). These adjustments to expected points added (EPA), in conjunction with opportunity-based metrics like total plays and drives, are then used to project adjusted points. The adjusted scores have been tested against actual scores for prior seasons and offer slightly better predictive ability, though their primary benefit is explanatory.

I’ve decided to make all the 2022 Adjusted Scores available to readers via Google Sheet. Check them out. I welcome your feedback.

LV (5-8) @ LA (4-9), 17-16 LA, Adjusted Score: 24-23 LV

The adjusted scores lean slightly toward the Raiders being the better team this game, and I think most watching would agree that they never should have allowed the Rams to be in position to steal it at the end. The Raiders were in a shell offensively the entire game, ending with a 40% dropback rate in comparison to the Rams at 69% (very nice).

Raiders coach Josh McDaniels probably believed the risk of passing, which Derek Carr displayed in flashing-red with an egregiously bad INT near the end zone (-4.4 EPA), was too great in an “easy” matchup with the Raiders running game steadily successful (50% rate). However, the Raiders reluctance to run on early downs (35%) in some ways turned the game into a higher variance affair, meaning success couldn’t happen as often in chunks, making drive outcomes dependent on the results of a handful of third down conversions.

The Raiders were only 5-13 converting third downs, three of those successes on the opening touchdown drive of the game. Outside of the desperation attempt at end game and the one field-goal sequence, all the Raiders second-half drives started with a run (netting -1 yards after penalty), and ended in punts on 4th & 1 or 4th & 2, once on the Rams side of the field.

The Raiders gave up nearly 8% in win probability by not going-for-it three different times the numbers indicated at least a 1% win probability gain, the second highest loss for a team this season. Ironically, McDaniels did make the call to go on 4th & 1 from the Rams 28-yard line in the third quarter when the numbers saw it as more of a toss-up. That type of field position is right in the range where field-goal percentage is high and you’re still far enough away from the end zone to make conversion-to-touchdown relatively unlikely. If anything, the McDaniels decisions hint that the Raiders don’t have model-based projections to guide fourth-down decisions, or McDaniels is simply ignoring them.

The 23-yard touchdown to Van Jefferson with only 15 seconds was the biggest play in win probability added. According to the nflfastR model, the Rams’ chance of winning jumped from a 14% to win all the way to 65%. It’s hard to capture win probability that late in the game, and I think the numbers from nflfastR likely were too conservative on both sides, meaning the Rams probably had a lower than 15% going into the play and higher than 65% coming out of it. Late-game models like this probably fall into the high bias-low variance category, which avoids the typical win probability model mistakes in overfitting, but also doesn’t capture the truth in these settings particularly well.

(FWIW, I reached out to Matt Davidow from Deck Prism, an expert for in-game win probability, and their model had the Rams win probability over 90% after the touchdown, which sounds right to me.)

Accuracy caveats asides, the late Rams score was second highest play in WPA this season by the model, behind only the miracle heave from P.J. Walker to D.J. Moore against the Falcons, which didn’t result in a win after a failed extra point and overtime.

Miracles aside, the Raiders are now out of the playoff hunt, suffering another come-from-behind loss that will stick with the team through the offseason. From the Rams perspective, they don’t gain anything concrete already having basically no chance at the postseason.

Yet, a competent performance for Baker Mayfield, on the shortest of prep time, is a good sign that perhaps McVay can work his magic again on another former No.1 pick. The Lions might be the biggest loser without even playing as owners of the Rams first-round pick. The pick is still in the No. 4 position, but its chance of moving up even further dropped significantly.