Bills 36, Texans 31: 37 of 46 Calls Had Houston, and Nobody Saw 67 Points

Buffalo won in Houston 36-31 in Week 1 of the 2026 season, and it settled 46 pre-game calls we had tracked on the game — 36 straight-up winner picks and 10 score predictions.

The field was close to one-sided. Of the 36 winner picks, 29 took Houston and 7 took Buffalo. Of the 10 score predictions, eight had the Texans winning. Add it up and 37 of 46 graded calls pointed the wrong way.

That matters more than a simple 29-7 count suggests, because Houston was the home team. Our default assumption in a game like this is that the home side wins, so a Texans pick earned no credit for insight — it graded as zero, the same as guessing. The seven people who took Buffalo on the road got the maximum: 100 accuracy and a skill mark of +200, meaning they beat the default by as wide a margin as one game allows.

The calls that landed

Richard Sherman had it twice. On September 5 he said of the Bills, "I think he's going to be a little rusty to start the season. I think Buffalo finds a way to get that game." Four days later he put a number on it: "I got Buffalo 20 to 14. I think Josh Allen overcomes his defense enough to, to pull out a win." Buffalo by six, against an actual margin of five — the single best-graded call on the game at 97.62 accuracy, worth +121.59 above the default.

Pete Prisco was right behind him with "Pick: Bills 23, Texans 20" — the Bills by three in a game that finished by five, graded 95.24.

The other Buffalo winner picks came from Danny Parkins ("Two chips on the Buffalo Bills. Josh Allen slays the Dragon that is the Houston Texans defense."), Albert Breer, AJ Hawk ("I gotta take the Buffalo Bills. On the road, Josh Allen... so give me the Bills."), Mike Kadlick, Kirk Herbstreit — who admitted he was partly rooting, not analyzing — and, from back on August 19, John Breech, who named this as the week Houston takes its first loss: "that will be enough to put Buffalo just barely over the top in this AFC showdown."

Breech then talked himself out of it. On September 8 he wrote "THE PICK: Texans 20-19 over Bills", and by September 13 he was taking Houston outright. Three graded calls on one game: one hit in August, two misses in September.

Nobody was close on the scoreboard

The wider miss wasn't the winner. It was the shape of the game. All 10 score predictions had this as a defensive grind — the highest total anyone forecast was 44. It finished 67. The average predicted total was 39.8, low by 27.2 points.

The actual margin was five, so the Texans-side score calls picked up partial credit for getting a one-score game right while missing the winner. Dubin and Dajani each published the same scoreline twice, and both versions graded identically.

The misses, named

Gregg Rosenthal leaned Houston: "The other one. In my heart is the Texans... We like the Texans, but they didn't quite crack our top four here." Rich Eisen picked the Texans as the likelier AFC front-runner by Monday morning. Kevin Wildes went with history — "Josh Allen loses yet again to the Texans in Houston" — and Joe Gibbs wrote that "it is difficult to envision a victory given their ugly recent past in Houston." Chad Millman put Houston in an underdog money-line parlay, and Judah Fortgang had Houston moneyline plus Stroud props built around the Texans winning through the air. Nick Wright was explicit back on September 1: "I do think they're gonna lose week one... even though they're a slight favorite, that's what's gonna happen."

Earliest of all, in May, John Granato said of Buffalo's visit: "Josh Allen's got no chance here."

What the records say

Sherman's hit is consistent with how he grades: 71.81 average accuracy on 11 graded score predictions and 71.43 on 15 winner picks, inside an overall record of 52.5 accuracy across 77 graded calls. His score work is the strongest part of his file, and this was the best of it.

Prisco's file is thinner — 17 graded calls in all, 15 of them score predictions averaging 65.72. One 95-accuracy hit is a good sign, not a track record yet.

Parkins is the interesting one. His winner picks sit at 60 percent accuracy across 25 graded calls but a skill mark of −18.42, meaning they've collectively done slightly worse than just taking the home team. A correct road call is exactly what that record needed. Breer, meanwhile, extends a solid run: 66.67 accuracy on 15 graded winner picks, 74.44 overall across 29 calls.

Rosenthal's miss barely moves his file — 373 graded calls, 61.54 accuracy overall, 58.68 on winner picks, plus a Pundit of the Week accuracy award last month. This was one lean, not a pattern.

What it touches next

We're tracking 25 unanswered win-total calls on Houston, nearly all of them double-digit, and the aggressive ones now need help: Lance Zierlein said in May he could see "two losses max", and Granato's 16-1 projection now requires Houston to win every remaining game. On the other side, 68 unanswered calls have Buffalo winning the AFC East, from Colin Cowherd, John Middlekauff, Jordan Dajani, Jared Dubin, Prisco and Breech among others. Week 1 did them no harm.

Both teams' running tallies are on the Texans and Bills pages, and the leaderboard has who's earning trust across everything we grade.