Bills earned a 23-6 win in Minnesota despite a 40 percent success rate
The snaps said Bills by 20 points, and one aborted snap in the first quarter did more than any pass the offense threw.

Air attackThe verdict
Played like that, the Buffalo Bills win more than 99% of the time. The snaps said Bills by 20.0 points, and the scoreboard said 23-6 over the Minnesota Vikings. That is a win earned, not borrowed.
The margin came from the defense, not the offense. Buffalo's offense was merely adequate by the numbers. Minnesota's was not.
What the score hid
EPA, or expected points added, measures how much each snap moves a team's scoring outlook. Buffalo's offense finished at -0.04 per snap, slightly below zero. Minnesota finished at -0.67, which is a long way below zero. Buffalo's success rate was 40%, Minnesota's 27%.
The luck sheet tilted the same way. Buffalo gained 0.8 luck points in this game and Minnesota lost 3.3. The Vikings were worse than the Bills and also got the worse of the bounces.
The swing play came in the first quarter with 5:30 on the clock. C.Wentz threw incomplete to J.Jefferson, Buffalo challenged the forward pass ruling, and the call was reversed. The result was an aborted-snap fumble at the Minnesota 43, recovered by G.Rousseau at the Minnesota 44. That one replay was worth 14 percentage points of win probability.
It is the turnover-worthy side of off-script football, and Minnesota's quarterback supplied it. Buffalo's offense produced no equivalent explosive play, and it did not need one.
The calls
Joe Brady's fourth-down work matched the model on all eight graded snaps. Five were punts, including fourth-and-21 from 47 yards out in the first quarter and fourth-and-23 from 98 in the second. Two were field goals, from the 21 in the first quarter and from the 6 on fourth-and-6 in the third.
The one aggressive call was the best one. With 10:15 left in the fourth quarter, Buffalo led comfortably and went on fourth-and-1 from the Minnesota 38. The offense converted. Brady took the model's answer on every snap, which is all that can be asked on a day like this.
The flags cost Buffalo more than the coaching did. Defensive pass interference in the third quarter at 13:26 gave Minnesota 19 yards and a 9-point swing in win probability. Intentional grounding in the first quarter, at 6:34, cost 13 yards and 6 points of win probability. Two false starts, one in the first quarter and one in the second, cost 5 yards and 3 points each. None of it changed the outcome.
The trend
Buffalo is 7-1, but the snaps support 5.6 deserved wins. That gap of 1.4 wins ranks second among the 32 teams, luckiest first. The Bills are 4-1 in one-score games. Of their 7.7 total luck points, 6.7 come from opponent kicking and 0.5 from fumble recoveries.
The underlying team is still strong. Buffalo's net EPA per snap is 0.135, fifth in the league. Brady's Chicken Index, the share of recommended go-for-it situations where a team punts or kicks, stands at 22%, 27th among the 32 teams ranked most timid first. He has gone on 7 of 9 fourth downs where the model said go, and the season's 38 graded fourth downs have cost 1.4 expected points, with one clear miss.
The penalty ledger is the soft spot. Buffalo has been flagged 68 times for 583 yards, and those flags have cost a net 10 percentage points of win probability across the season.
Buffalo plays the New York Jets on the road in Week 10, Sunday, Nov. 15, at 1:00 PM ET on CBS.