Dear readers, we’ve done it. In a week full of unexpected happenings, add another to the list: I’ve finally actually gotten the weekly review out by Tuesday morning.
Please clap.
Anyway, below you can find a series of charts for each matchup that cover what happened in the game from both a real-life football and fantasy football perspective. The analysis will be stat-heavy and lean on expected points added (EPA) and success rate (the percentage of plays that produce positive EPA), thanks to data from the indispensable nflverse.
We’ll start as always with the Monday and Sunday night games, since those should be freshest in our minds, before working back to the Thursday nighter, Europe game and finally the Sunday afternoon slate.
However, speaking of the newsletter schedule, I probably won’t have time to get a full Week 6 preview out in time this week, so the next bit of football numbers goodness will probably be the Week 6 review. I also have a baseball playoffs piece in the works that will probably be ready next week.
Note that all rankings, including the listed percentiles, do not include the results of the Monday Night Football game, with the exception of the numbers shown in the Monday Night Football review itself.
Bills 37, Rams 34 (OT, Monday Night Football)
The Rams lost a whopping seven expected points on their field goal attempts alone, as Harrison Mevis was 0-for-2, with a miss from 50 yards and a blocked attempt on a 48-yarder.
Puka Nacua caught 17 of his 24 targets for 138 yards and a touchdown.
Ravens 30, Falcons 23 (OT, Sunday Night Football)
Both teams turned the ball over once.
Baltimore’s was more costly from an expected points added perspective; Mark Andrews’ fumble that was returned 36 yards for a touchdown in the first quarter cost the Ravens about 7 expected points.
Atlanta’s was more costly from a win probability added perspective; Tua Tagovailoa’s interception in overtime took the Falcons from having about a 50% chance of winning the game to a zero percent chance.
Michael Penix produced a healthy 0.22 EPA per dropback in his 35 dropbacks before leaving the game with a rib injury. This was on the back of explosiveness more than efficiency, as while he averaged 12.6 yards per completion,1 he also had just a 43% success rate and a minus-5% completion percentage over expectation (CPOE).2
Huntley was solid filling in for the injured Lamar Jackson, completing 24 of his 32 passes for 201 yards and two touchdowns without any interceptions. He also added 20 yards on the ground on six attempts (three of which were designed rushes), including a 1-yard touchdown rush in overtime that proved to be the game winner (more on that in a moment).
It wasn’t a perfect night, as the Ravens passing game almost completely lacked explosiveness3 and Huntley was sacked three times.4 Still, as their Sunday night opponent can attest, things can certainly go way worse for a backup quarterback.
It was a boom/bust kind of night for Bijan Robinson, who averaged 5.1 yards per carry (86 yards on 17 runs), but only managed a 29% success rate.
Who would have thought that in a game featuring Robinson and Derrick Henry, Huntley would have the game’s lone rushing touchdown?
The two stars became the latest backs who failed to convert near the goal line, as Henry had three goal-to-go carries and Robinson had two.
Buccaneers 24, Cowboys 16 (Thursday Night Football)
Dallas’ net turnover EPA was actually below minus-14, as the turnover EPA shown in the chart is only from plays from scrimmage; the Cowboys lost a fumble on a kickoff (to go along with two Dak Prescott interceptions).
Dallas had a five percentage-point edge in success rate (51% to 46%), but the overwhelming majority of their positive plays were clustered on first and second down.
Javonte Williams had 12 of Dallas’ 13 designed runs on Thursday and now has a 77.1% share for the season; only Jahmyr Gibbs, Kenneth Walker and Derrick Henry have higher rates of their team’s designed runs.
Emeka Egbuka’s final line roughly matches what the expected stats dictated, but he did this in a deceptive way. He had nine targets with a 7.8 average depth of target (aDOT); he matched his receptions expectation (6), but underperformed in the receiving yardage (38 actual, 67.3 expected) and touchdown (0 actual, 0.7 expected) departments.5
Egbuka salvaged his fantasy night with a 14-yard touchdown rush (on a play that was expected to gain 3.8 yards with a roughly 10% chance of scoring a touchdown).
Jaguars 19, Eagles 17 (London)
Somehow this was not the strangest ending to a game featuring an NFC East team this week…but it was still pretty wild.
I’d say the model is actually overstating how likely the late Jacksonville comeback was, and the ESPN win probability model agreed; the latter model had Philadelphia’s chances of winning at 99.9%.
It’s a good reminder that 1-in-1,000 events can happen, a reminder I could have used on Sunday. I actually flipped this game off after Philadelphia kicked its field goal to go ahead 17-16 since the 1 p.m. games had already begun to kick off. Oops.
The Eagles had the edge on all the high-leverage snaps — they won the turnover battle, were a smidgen better on late downs, and dominated the red zone — and still lost.
Philadelphia scored on its lone red zone possession, while Jacksonville was held out of the end zone on all three of its trips inside the 20 (and was stopped on fourth down on two of them).
The Eagles are neck-and-neck with the Chargers in the race for most disappointing offense, but they are also in contention for just the flat-out worst offense, period, setting expectations aside.
Philadelphia is last in the league in success rate (35%) and second to last in EPA per play (minus-0.12, ahead of only the Chargers). Personnel losses have not helped, but the Eagles offense was also not firing on all cylinders in the early-season games that Saquon Barkley, DeVonta Smith and Lane Johnson did appear in.
According to the on-off tool at raycarp.com, in the 78 snaps that trio shared this season, the Eagles managed a 36% success rate and averaged minus-0.06 EPA per play. While on/off splits are far from the end-all, be-all, they do serve as further evidence that the team’s issues go deeper than just injuries and Johnson’s retirement.
Parker Washington turned his nine targets into six catches for 124 yards and two touchdowns. The yardage and touchdown totals were well above expectation (65.6 and 0.4, respectively).5
Giants 27, Commanders 27
Note that the chart above is referring to Washington’s win probability in a literal sense, in that a tie and a loss both have 0% win probability. I usually like to think of the inverse of these charts as the other team’s win probability, but this game served as a reminder that this technically isn’t the case.
This might have featured the most bizarre ending we’ve ever seen, with Cam Skattebo fumbling away the Giants’ chances to win in overtime, only for Jayden Daniels to inexplicably take a sack and let the clock run out on the game’s final play.
Washington’s colossal bungling of its final drive has somewhat overshadowed its colossal bungling of the fourth quarter of this game. The Commanders took a 14-point lead with 11:05 left in regulation, before surrendering a 75-yard touchdown drive, going three-and-out, surrendering a 72-yard touchdown drive, and being unable to break the tie despite getting the ball back with 1:36 left in the fourth quarter and two timeouts.
Aside from the miserable overtime execution, the high-leverage downs were the story here: the Giants were 10-for-19 on late downs and scored touchdowns on all three red zone trips, while the Commanders were 4-for-12 on late downs and failed to find the end zone on their lone red zone possession.
For the season, Washington has the sixth-largest gap between its early down success rate (39%) and its late down success rate (35%). The median difference is minus-1%, as the league-wide average success rate is higher on late downs.
Jayden Daniels was perfect on throws in the intermediate range (10 to 19 air yards) but had less success on shorter and deeper throws.
The Commanders managed a subpar 41% success rate on throws shorter than 10 yards,7 while Daniels was 0-for-3 on throws at least 20 yards down the field.
Malik Nabers and Isaiah Likely combined for 63.5% of the Giants’ targets.
Nabers had 11 catches for 121 yards and a touchdown, yet he quite possibly left fantasy points on the board; he had 20 targets, including four in the end zone, with a 12.2 aDOT.
Likely, meanwhile, beat his expectations, catching 11 of 13 targets (two in the end zone) for 94 yards and two touchdowns.
Quarterbacks are meant to be excluded from the fantasy charts here, but J.J. McCarthy shows up since he did not have any official pass attempts,6 but did have a pair of designed runs.
49ers 32, Seahawks 29
San Francisco’s muffed punt is included in the special teams EPA row but not the turnovers row.
Jason Myers was 5-for-5 on field goals for Seattle, with a long make of 50 yards.
Steelers 34, Colts 22
Since the start of last season, the Colts have been the league’s third-best offense in terms of EPA per play in the red zone. You would not have known it watching them on Sunday, as Indianapolis scored only one touchdown on three red zone trips against the Steelers.
Aaron Rodgers kept things short on Sunday after attempting to stretch defenses for big chunks of the early season. He only had a 4.6 aDOT against the Colts.
Jaylen Warren had an 80-yard touchdown catch and a 50-yard touchdown run.
Jonathan Taylor failed to score on four goal-to-go carries and only had a 20% success rate. His designed rush share was also just 62.5%; he had 15 carries, and Seth McGowan had five, while Daniel Jones and Tyler Warren added three and one, respectively (Jones and Warren had a carry apiece at the goal line).
Laquon Treadwell was third on the team in targets with eight (behind Warren and Keenan Allen who had nine apiece), but could only manage two catches for 34 yards.
Packers 31, Bears 7
Patriots 45, Raiders 17
Drake Maye had a monster day, completing 26 of 34 passes for 374 yards and four touchdowns with no interceptions.
Lions 45, Cardinals 21
Amon-Ra St. Brown had the chance for a much bigger game, as he was targeted 11 times (good for a 38% target share), with two coming in the end zone. However, he could only turn them into six scoreless catches for 69 yards.
Broncos 23, Chargers 17
The injury-riddled Chargers had a win probability north of 80% in the fourth quarter, before surrendering 13 consecutive points to earn another defeat.
Jim Harbaugh was hard on himself for attempting a 4th-and-1 in the fourth quarter with his team up seven, but based on the data, this was a perfectly defensible decision. Teams on average are more likely than not to convert when they need a yard, and the rewards seemed worth the risk.
Vikings 23, Saints 13
The turnover EPA row does not include the fumble New Orleans lost on a kickoff, but the damage is accounted for in the special teams row.
Minnesota lost more than 10 expected points on a fumble from Myles Price that was returned for a touchdown.
Jets 17, Browns 12
Bengals 31, Dolphins 24
This was the epitome of a game that was not as close as the scoreboard suggests; Miami’s win probability was below 20% for almost all of the second half.
Texans 26, Titans 3
1 For context, Patrick Mahomes and Bryce Young came into Sunday tied for fourth in the league with a 12.6 yards per completion average.
2 The completion probability model looks at a variety of factors, including how far a pass was thrown and field position, to determine the probability of a pass being completed. CPOE is calculated by comparing the rate of passes a player actually completed versus the rate the model predicted them to complete.
3 Baltimore had just a single pass gain at least 20 yards, while Huntley averaged a meager 8.4 yards per completion.
4 Zay Flowers also took a sack on a failed trick play.
5 I use my own expected yardage and expected touchdown models, and the nflverse completion probability model for expected receptions. I combine these outputs to calculate expected fantasy points. You can find the code for the touchdown and yardage models, in addition to the code I use for this analysis, in this GitHub repository.
6 He threw the ball twice on Sunday but both plays were negated due to penalty.
7 The NFL average is 47%, not counting the Monday night game.