Bills 41, Lions 31: The Field Took Buffalo 16-4 in a Predictable Shootout
Buffalo beat Detroit 41-31 on Thursday night in Week 2, settling 20 calls we had on the books — nine straight winner picks and eleven score predictions.
The field was close to one-way. Sixteen of the 20 took Buffalo; four took Detroit.

Picking the home team
Every winner pick that took Buffalo graded a clean 100 for accuracy and zero for skill. Buffalo was the home team, and our skill number measures whether you beat the dumbest possible strategy of always picking the host. Nobody who took the Bills beat it and nobody who took Detroit got paid for the courage either (because the road team lost.)
Richard Sherman alone accounts for three of those calls, made across three weeks and never wavering. Back on September 1: *"I have a hard time believing Buffalo is going to lose the first game in their brand new stadium."* On September 5: *"Your Detroit Lions, again, that's a tough game to go to Buffalo in their new stadium and get that win. I think Buffalo finds a way in that game."* And on the morning of the game: *"I got the Buffalo Bills in this one. I'm sorry, Detroit."* Three for three, and three skill deltas of zero.
Joe Marino was more specific about the shape of it — the Bills win and go 2-0 in a shootout with Goff and Allen combining for 600-plus passing yards. The winner half is graded and correct. John Breech had it locked in back on August 19, in a piece projecting when each team takes its first loss: *"it's almost impossible to see the Bills losing the first game at their new stadium."*
Separation through score estimates
Eleven score predictions rode on this game, and unlike the winner picks, a good score call does earn credit above the baseline. The margin ended up being 10.

| Pundit | Prediction | Margin called | Accuracy | Credit vs. baseline |
|---|---|---|---|---|
| John Breech (×2) | Bills 38-31 | 7 | 92.86 | +26.08 |
| Walter Cherepinsky | Bills 31-24 | 7 | 92.86 | +26.08 |
| Jordan Dajani (×2) | Bills 36-30 | 6 | 90.48 | +21.46 |
| Pete Prisco | Bills 36-30 | 6 | 90.48 | +21.46 |
| Tyler Sullivan | Bills 31-27 | 4 | 85.71 | +12.23 |
| Jared Dubin (×2) | Bills 35-31 | 4 | 85.71 | +12.23 |
| Jared Dubin | Bills 35-32 | 3 | 83.33 | +7.61 |
| Tony Fusco | Lions 87-12 | 75 (Detroit) | 0 | −154.06 |
Breech's 38-31, published September 15 and repeated two days later, and Cherepinsky's 31-24 were the best-graded calls of the night at 92.86. Dubin's 35-32 got the winner and the shootout right but called a three-point game that finished by ten — still an 83.33, the softest grade among the correct scores. Fusco's *"I like the Lions to go big here. Lions 87, Bills 12"* comes from his satirical predictions column; it grades like satire, at zero.
The misses
Three winner picks took Detroit, and each scored zero accuracy and roughly −194 on skill — worse than defaulting to the home team.
Colin Cowherd was the bluntest: *"I like the Lions to cover. I like the Lions to win."* Kevin Wildes had made his back on September 4 as part of a full-season grid — *"A lot of this is based on, hey, I think Dan Campbell will have the Lions ready to play... I believe in the Lions."* And Tyler Sullivan landed on both sides of the same game: his CBS Sports upset-alert column flagged Detroit (+4.5) as a live dog, while his straight score pick that same day was Bills 31-27. One graded 85.71; the other graded zero.
What it did to the records
Cowherd was the most-impacted. He is one of the few voices we have an deep record on — 451 graded calls, 55.75 accuracy, a MissedTakes Score of 7.67 — and his game-outcome work is strong: 70.65 accuracy over 73 picks, with a good skill credit (+10.85).
Breech is the opposite. His overall record sits at 47.99 accuracy and a MissedTakes Score of −9.75 over 42 calls, and his score predictions have run below the baseline (−22.01 over 20 calls). Thursday was his best night on the site so far.
Sherman's game-outcome record is now 82.08 accuracy over 19 picks with a positive skill score (+10.08), though three of those 19 came from this one game. Wildes has nine graded winner picks at 25 accuracy — thin enough to call a lean, not a record. Cherepinsky's only two graded calls are the two from this game, so his 96.43 accuracy is a single good night, nothing more.
What the result touches next
Detroit's loss doesn't undo much yet — Stephen Oh's post-Week 1 simulation still has the Lions as NFC North champions and the Bills as the AFC's No. 2 seed, and Garrett Podell argued this week that Buffalo has as good a shot as any AFC team to reach Super Bowl LXI. The call this hurts most is Wildes's preseason grid, which had the Bills missing the playoffs entirely and the Patriots beating the Lions in the Super Bowl; Buffalo is 2-0.
Full call-by-call histories: Buffalo Bills, Detroit Lions, and the leaderboard.