Updated · market price
Sunday, September 13, 2026 · NRG Stadium, Houston, TX
Quick Answer
Final (Sunday, September 13, 2026): Buffalo Bills 36, Houston Texans 31 — NFL Week 1, 2026. Our pre-game model had the Texans by 0.5 (53% to win); the Bills won, so that call missed. The graded recap below covers how it happened and how each player call on our board resolved.
Settled
Buffalo Bills 36, Houston Texans 31Final
Model call: MISSOur pre-game model had the Texans by 0.5 (53% to win). The Bills won — that call missed.
Closing-line value on the Texans: -3c — we archived the Texans at 51c on Sep 7; the last Kalshi price before kickoff was 48c on Sep 12. The market moved against our side before kickoff.
Next: NFL Week 2 predictions →See how the model has done vs Kalshi →
No actionable edge on this game — 1 market priced and every one inside our tolerance.
Edge is model probability minus the market's implied probability, in percentage points. One published number per market: where our sources disagreed we publish the most conservative defensible figure, and where they disagreed by more than 15 points the row is a conflict and carries no stake. ¼ Kelly is the share of a bankroll quarter-Kelly would put on the row at the published price — click it to size against your own bankroll on our Kelly calculator, or call kelly_size from our MCP server inside your own agent.
A true pick — Buffalo -1.5 open and close, moneylines within two cents of each other in both directions — and the only thing the line did all week was take a point off the total. The teams then produced sixty-seven. Josh Allen threw for 334 at 11.5 an attempt and scored twice on the ground, and Houston lost two C.J. Stroud fumbles.
Bills 36—Texans 31
Final✗ Call missedThe model missed the side and the position lost. We had Houston at 52.7% against a Kalshi price of 47.5c, a 5.2-point gap, tiered MODERATE, published Houston at 49.5c. It closed at 48.5c and Houston lost by five. That is the smallest edge we published on the slate and it is instructive precisely because it is small: on a genuine coin-flip the model and the market were a nickel apart and the market was slightly closer. The total is the interesting part again. DraftKings opened 45.5 and closed 44.5, moving down a point, and the game landed twenty-two and a half over on 7.9 yards per play from Buffalo and four fourth-quarter lead changes. That is now the second game on the slate where the closing total move pointed the wrong way — the other was Atlanta at Pittsburgh in reverse — and we published no total position on either. Our total model is not the thing that lost money in Week 1; it is the thing that was not in the room.
DraftKings via ESPN, one book. Not the Kalshi price — that block reads our own tape.
The spread did not move. The total fell 1 to 44.5; the game landed on 67.
Q2
Allen ran in from twenty-one and then from one inside the second quarter, and threw a forty-three-yard touchdown to DJ Moore with ninety seconds left in the half. Buffalo scored twenty-one in the quarter. He finished with 334 passing yards on twenty-nine attempts and two rushing scores on six carries.
Q3
Kincaid caught five for 130 with a long of forty. Our board carried MODERATE UNDERs on Kincaid at 39.5 receiving yards and 3.5 receptions; he cleared the yardage line by more than ninety and the reception line by one. DJ Moore added five for a hundred and the score.
Q4
Houston went ahead 28-27 on a seven-yard throw to David Montgomery, Buffalo retook it on a Tyler Bass field goal, Ka’imi Fairbairn answered from fifty-eight for 31-30, and Joshua Palmer caught a thirty-four-yard touchdown with 1:36 left. The two-point try failed, leaving it a five-point game and the onside as the only route back.
Q4
Houston held the ball for 36:17, converted seven of sixteen third downs and both fourth downs, and gained 381 yards. Stroud lost two fumbles; Buffalo turned it over zero times. Houston also took seven penalties for 106 yards, the most penalty yardage of any team in the early window.
Reporting verified 2026-09-13. Further sources: ESPN box score — Bills at Texans · ESPN game page — Bills at Texans · ESPN — Week 1 highlights: best plays, moments, touchdowns
Houston's front seven put Josh Allen on the ground eight times the last time these teams played. Buffalo arrives with a new head coach, two new coordinators, a rebuilt run defense and no fullback — but also the running back who already beat this same Houston front once. The page above has this as one of the closest games of the week, and the trenches are why.
Texans pass rush · Bills offensive line
Anderson and Hunter did most of the damage the last time these teams met, and Houston brings the same pair back. Buffalo returns four of five starters up front but is breaking in a new offensive coordinator in Pete Carmichael Jr., and per beat reporting will play without a fullback for the first time in Allen's career — one more moving part in pass protection against the duo that already solved it once.
Texans offensive line · Bills edge rush
Houston is still settling its own interior. Left guard Keylan Rutledge only moved off center into his natural spot in the final week of camp, and free-agent addition Evan Brown won the center job over last season's starter. Buffalo counters with Rousseau and a big free-agent signing in Chubb, replacing the departed Joey Bosa and A.J. Epenesa. Nobody has graded this specific pairing, so it is a read, not a lock, in either direction.
Bills run defense · Texans backfield
Buffalo was one of the worst run defenses in the league a year ago, both on yards per carry and on explosive runs allowed. First-year coordinator Jim Leonhard is installing a new scheme in real time rather than after a full offseason of reps. Montgomery projects as Houston's early-down back against a unit that graded this poorly at exactly what he does.
Bills backfield · Texans run defense
Whatever Houston's defense did to Allen in November, Cook found room against it — a 1,000-yard season and a 45-yard touchdown run in that same game. Houston's run-defense grade in isolation, separate from its overall defensive number, was not independently sourced for this read, so treat this as a favorable history note rather than a settled ranking.
Bills secondary · Texans receivers
Christian Benford is the only returning starter from Buffalo's 2025 cornerback group; Maxwell Hairston and Dee Alford fill in around him. Across from them: Collins as the clear No. 1, trade acquisition Boutte on the other side and Jaylin Noel in the slot. Standard alignment puts Houston's more proven trio of Derek Stingley Jr., Kamari Lassiter and Jalen Pitre on Buffalo's receivers and Buffalo's thinner group on Collins. No outlet has confirmed shadow assignments, so that is inference.
Bills QB · Texans QB
Allen has never won a start in Houston. Stroud's status needed a genuine double-check, because search results still surface a concussion-and-Davis-Mills item — it dates to the November 2025 meeting, not this week. Houston's Sept 2 depth chart and its Sept 9 injury report both list Stroud as the healthy starter; he was rested, not hurt, in the preseason opener.
Source: CBS Sports — Week 1 expert picks
Bills coaching staff
Joe Brady is a first-year head coach, and both coordinators — Pete Carmichael Jr. on offense and Jim Leonhard on defense — arrived this offseason out of the Sean Payton tree in Denver. Several published previews cite exactly that, new schemes on both sides of Buffalo's ball, as their reason to expect a looser game than the ratings suggest. That is their read, not a number of ours.
Line shopping
DraftKings, BetMGM and FanDuel all had Buffalo favored by roughly a point to a point and a half as of Sept 7–9. Kalshi and a separate Covers.com read had it closer to a pick'em with Houston a hair ahead. The number on this page sits inside that range, which is the honest story: this is one of the tightest reads on the board, not a game with a hidden number nobody else sees.
Source: OddsShopper — Bills vs Texans prediction-market odds
Reporting verified 2026-09-11. Further sources: Bills.com — Week 1 injury report vs Texans · SI — Bills projected Week 1 lineup · SI — Texans updated depth chart · Houston Texans — Anderson, Hunter and the blueprint Buffalo has to answer · Yahoo Sports — grading the Texans' 2026 offensive line · Covers.com — Bills vs Texans early leans
BUF
+3.8
HOU
+2.0
Our number and the market agree on this one — the gaps are elsewhere on the board. Get the full Week 1 card free, every game priced against our model.
No spam. Unsubscribe anytime.
Bills
47%
model win prob
Texans
53%
model win prob
Run this matchup yourself → Full PWR board for all 32 → Player prop gaps for this game →
52%
Madden 27
48%
Madden 27
Go Pro · 26% off your first year
Pro is the layer around the free game edges: model win probability against the live Kalshi price on every market, edge alerts in Discord the moment a gap opens, and the full win-total, championship and projection boards.
$111 for the first year, then $150/yr. Code applies automatically. Ends Dec 31. Monthly is $14.99/mo.
Every signal graded in public, losers left on the board. See the record →
Our margin distribution against every Kalshi spread strike listed for this game. A single line says we lean one way; the ladder says whether that is about the middle of the distribution or its tail.
Our number is published at the called strike only; the market curves are complete.
| Rung | Market | Model | Gap | |||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Texans −0.5 | — | 🔒 | — | not listed | ||||||||||||||||||||
| Texans −1.5 | — | 🔒 | — | not listed | ||||||||||||||||||||
| Texans −2.5 | — | 🔒 | — | not listed | ||||||||||||||||||||
| Texans −3.5 | — | 🔒 | — | not listed | ||||||||||||||||||||
| Texans −4.5 | — | 🔒 | — | not listed | ||||||||||||||||||||
| Texans −6.5 | — | 🔒 | — | not listed | ||||||||||||||||||||
| Texans −7.5 | — | outside range | — | not listed | ||||||||||||||||||||
4 more rungsTexans −9.5 → Texans −14.5 · outside our published band, market only
| ||||||||||||||||||||||||
Rung-by-rung numbers are a Pro feature. Unlock Pro →
Gold is our distribution, parchment is the Kalshi contract at each strike; the band between them is green where our probability is higher and purple where it is lower. We publish our number only where we have measured it to be calibrated, between 30% and 70%; rungs outside that range show the market only. A contract pays $1 if it settles YES, so a price in cents is an implied probability. Prediction markets carry risk; contracts can settle at $0.
10,000 simulated seasons on our preseason power ratings — real NFL tiebreakers, real 2026 schedule, no injuries or market prices. The season sim fits its own home-field (2.0 points, from played games); the game model above uses +2.5. How it works →
The better team here is the visitor, and it still is not favored. The Bills rate +3.8 (4th of 32) against +2.0 for the Texans (9th) — but 2.5 points of home field is larger than the 1.8-point gap between them, so the model lands at 53% for the Texans anyway. That is the whole story of this line: venue, not talent. It rates as the 4th-closest of the 16 games in Week 1, and a game this tight is decided by the price, not the model.
Season-long, the model can barely split them: 9.3 projected wins for the Texans against 10.1 for the Bills, favored in 12 and 13 of their 17 respectively. Two teams heading for the same neighborhood in the standings tend to produce exactly this kind of line. This game sits 7th-hardest on the Texans' schedule and 4th-hardest on the Bills'.
The sharpest unit gap in the game runs one way: the Bills offense grades +4.1 (2nd of 32) against a Texans defense at +2.6 (1st) — 1.5 points of separation on our scale, and the single biggest reason the line sits where it does. Going the other direction, the Texans offense (-1.3, 25th) meets the Bills defense (-0.7, 16th), a 0.6-point gap.
Recent history, for context only: 12 regular-season meetings since 2002, 7-5 in the Texans' favour, plus 1 playoff game. They last met in 2025: Texans won 23-19 in November. Over their last 4 meetings the Texans are 3-1. It is not an input to the line — rosters turn over faster than series records do.
Calibration check, because a near-even number is the easiest kind to overtrust: across 3266 walk-forward games (2013-2024) the model gave a side 50-55% in 870 of them, and those sides won 52%. The Texans at 53% are in that band. A margin that thin is why we would rather wait for a price than call it a lean.
Kalshi's KXNFLGAME-26SEP13BUFHOU contract implies 50% for the Texans against our 53% — a 2.6-point gap. That clears our 5-point threshold, which flags the Texans as the value side.
Prediction markets are not sportsbooks — you trade a contract that settles at $1 or $0, not a wager against a house line. Informational only; trade responsibly.
This game has been played, so nothing above is a live market. Our pre-game model number was 53% for the Texans. Those are kept as the record of what the model said before kickoff, not as a quote you can act on — the contract has settled at $1 or $0.
2 venues price this game on their own exchange; the rest resell the Kalshi contract or list it without a price we can quote. Home-side price shown.
| Venue | Texans | Trade |
|---|---|---|
| Kalshi | 50% 50c | Trade on Kalshi → |
| Novig | 48% 48c | Open Novig → |
| Polymarket | no captured price for this game — international book | Sign-up offer → |
| Robinhoodsame Kalshi contract | 50% 50c | Open Robinhood → |
| Coinbasesame Kalshi contract | 50% 50c | Open Coinbase → |
| DraftKings Predictions | price not available — sign-up offer → | Sign-up offer → |
| FanDuel Predicts | price not available — sign-up offer → | Sign-up offer → |
| Fanatics Markets | price not available — sign-up offer → | Sign-up offer → |
A price under a venue name is that venue’s own; Robinhood and Coinbase list the same Kalshi contract. DraftKings Predictions, FanDuel Predicts and Fanatics Markets run on exchanges we have no price feed for, so they carry an offer and no number. We earn a referral credit or commission when you sign up through these links; it never changes the price or the model. Disclosure.
Buffalo Bills 36, Houston Texans 31 — final, played Sunday, September 13, 2026; our pre-game model had the Texans by 0.5 (53% to win), and the Bills won, so that call missed. The full graded recap, with every player call resolved, is on this page.
Before kickoff our power-rating model favored the Texans by 0.5 points, a 53% win probability over the Bills. That number is kept on this page as the record of the call; Kalshi's KXNFLGAME-26SEP13BUFHOU market has since settled, so no price on it is live.
Because the Texans (+2.0, 9th) and Bills (+3.8, 4th) are only 1.8 rating points apart. The 2.5-point home-field adjustment is larger than the gap between the teams, so home field is effectively the entire margin and the split lands at 53%/47%. In games like this the model is not the deciding input — the contract price is.
The contract is KXNFLGAME-26SEP13BUFHOU on Kalshi, a CFTC-regulated exchange open to traders in all 50 states. Texans YES is quoted around 50c against our 53% model number. There is no point spread — you buy a yes or no contract on a team and it settles at $1 or $0.
Buffalo Bills visit the Houston Texans on Sunday, September 13, 2026 — a Sunday game in NFL Week 1, at NRG Stadium, Houston, TX.
Across all 17 games the model projects the Texans at 9.3 wins (favored in 12) and the Bills at 10.1 (favored in 13). This game ranks 7th-hardest on the Texans' schedule and 4th-hardest on the Bills', by model win probability.
Kalshi prices this game as event KXNFLGAME-26SEP13BUFHOU. The Texans YES contract sits around 50c, which is an implied 50% — against our model's 53%. That is a 2.6pp gap, and a contract price in cents is already a probability, so no odds conversion is needed.
This game has been played, so the Week 1 market KXNFLGAME-26SEP13BUFHOU has settled — there is no position left to take on it. While it was open it traded as an event contract rather than a point spread: Kalshi is a CFTC-regulated exchange open to traders in all 50 states, you bought a YES or NO contract on a team between 1c and 99c, and it settled at $1 or $0, so the price paid was the break-even probability. Kalshi lists the next week's games on the same terms.
52% of the time. Over 3266 walk-forward games from 2013-2024 — trained only on prior seasons, no market line as an input — the model put a team in the 50-55% range 870 times, and those teams won 52% of them. The Texans at 53% over the Bills sit in that band, off a 0.5-point model line. Straight-up across all 3266 graded games the model picked winners at 59.6%, with a Brier score of 0.236 against 0.25 for a coin flip.
Our model puts it at 53% against a market price implying 50%, an edge of +2.6 points. Model 53% vs market 50% on HOU. We pass — no position.
Every Thursday before kickoff: the full Week 1 card, market price vs our number. Free.
No spam. Unsubscribe anytime.
The engine behind these numbers is callable from Claude, ChatGPT or Grok through our MCP server — every Kalshi prop this week priced venue by venue, spread and win-total ladders against the whole distribution, correlation-aware same-game combos, and the weekly game edges. The free tools answer with no key and no account.
See the in-season NFL tools →Prediction markets carry risk. Contracts can settle at $0. Informational only — not financial advice. Trade responsibly.
