Vikings lose by 17 and the snaps say it should have been worse
Minnesota managed a 27 percent success rate and minus-0.67 EPA per snap against Buffalo, and the snaps put the gap at 20 points.

The verdict
Played like that, the Vikings win zero percent of the time. The snaps gave Minnesota less than a 1 percent chance against the Bills, and the engine's deserved margin was Buffalo by 20.0 points. The 23-6 final was close to the truth. This was a deserved loss.
The offense never produced a rhythm. Nothing in the game's structure, from the early turnover to the one-sided penalty count, explains it away.
What the score hid
The Vikings averaged minus-0.67 EPA per snap. EPA, expected points added, measures how much each play moves the scoreboard against what a typical team would get from that spot. Buffalo finished at minus-0.04. Neither offense was good, but one was far worse.
Success rate tells the same story: 27 percent for Minnesota, 40 percent for Buffalo. Fewer than three snaps in ten kept the Vikings on schedule. An offense built on open receivers after the catch needs those early downs, and it did not get them.
Luck points were no help either: minus-3.3 for Minnesota, plus-0.8 for Buffalo. The swing play came in the first quarter at 5:30. Carson Wentz's pass to J.Jefferson was ruled incomplete, then Buffalo challenged the forward pass ruling and it was reversed to an aborted-snap fumble. G.Rousseau recovered at the Minnesota 44, and the play moved win probability by 14 points.
The calls
The flags went Minnesota's way. Buffalo drew a 19-yard pass interference penalty in the third quarter at 13:26, worth 9 points of win probability, plus intentional grounding for 13 yards in the first quarter and two false starts. The Vikings still scored six points.
On fourth down, the model disagreed with the sideline twice. In the first quarter at 1:08, Minnesota punted on fourth-and-5 from midfield, and the model said go, costing 0.3 expected points. In the second quarter at 4:06, the Vikings kicked a field goal on fourth-and-1 from the Buffalo 25, and the model said go, costing 0.6 expected points. The other four fourth-down calls matched the model.
The head coach's season record on these decisions is cautious. Minnesota has graded 54 fourth downs, and the model recommended going on 13 of them. The Vikings went on 5. That gives a Chicken Index of 62 percent, ninth most timid of 32 teams, with 5.6 expected points lost and 2 clear misses. Sunday's two calls were small leans, not disasters, but they fit a pattern.
The trend
The record is 6-2. The snaps say 4.7 deserved wins, a gap of 1.3 wins, which ranks third among the 32 teams for luckiest. Minnesota is 4-1 in one-score games, and its total luck points stand at 5.5, including 4 from opponent kicking.
The net EPA per snap is 0.008, 15th in the league, which is a middle-of-the-pack offense and defense with a record that runs ahead of it. Those numbers are due to come down unless the offense starts winning early downs.
The Vikings play at Green Bay in Week 10, Sunday, Nov. 15 at 1:00 PM ET on Fox.