NFL Week 4 2026 Results: Our Predictions Graded
Updated · market prices
This week's card: NFL Week 4 picks: the best play from every Sunday game →
Quick Answer
As of October 5, 2026, Week 4 is final. The model went 9–7 straight up and 3–6 on published positions (Kalshi vs book consensus) — every game below is graded against its final score.
Each game links to its own page with the final score and how our call held up; the full record lives on the track record.
This week: NFL Week 8 predictions.
Week 4 straight-up picks
The model's winner in all 16 games, graded below against the final scores. Each row shows the pre-game call.
- Steelers over Browns · pre-game call59%—
- Colts over Commanders · pre-game call65%—
- Ravens over Titans · pre-game call85%—
- Bills over Patriots · pre-game call73%—
- Bears over Jets · pre-game call62%—
- Bengals over Jaguars · pre-game call59%—
- Texans over Cowboys · pre-game call60%—
- Cardinals over Giants · pre-game call59%—
- Rams over Eagles · pre-game call62%—
- Packers over Buccaneers · pre-game call60%—
- Vikings over Dolphins · pre-game call80%—
- Chiefs over Raiders · pre-game call65%—
- Seahawks over Chargers · pre-game call73%—
- 49ers over Broncos · pre-game call60%—
- Lions over Panthers · pre-game call62%—
- Saints over Falcons · pre-game callUpset55%11.5¢
Winner = the side with the higher published win probability. Once Kalshi prices a game that probability is market-anchored — it is the market’s number, which is why the Model and Kalshi columns sit within a point of each other and why no upset is labelled. Before a market opens it is the model’s own read. The model’s separate opinion is the lean, in points, graded on its own and never published as an edge. Kalshi = the mid price of the winner’s contract, in cents. A straight-up call is not the play — the play on each game is the edge side on its matchup page.
Week 4 in numbers
- 9–7 straight up for the model, every game graded below.
- 7 upsets by closing line among the 16 games recapped, led by the Patriots at +275 (+7): Patriots 29, Bills 26.
- The line moved toward the winner in 8 of 16, toward the loser in 6, flat in 2; season to date, toward the winner in 49 of 106, toward the loser in 42, flat in 15.
Biggest moves of Week 4
- Cardinals at Giants: moved 9.5 toward the Cardinals (Giants −7 → +2.5); Giants won by 12 anywaytoward loser
- Colts at Commanders: moved 6 toward the Colts (Commanders −1.5 → +4.5); Colts won by 17toward winner
- Jets at Bears: moved 5 toward the Jets (Bears −8.5 → −3.5); Bears won by 11 anywaytoward loser
One line source: DraftKings, via ESPN — the opening and closing spread on each game, not a consensus of every platform. Every 2026 upset →
Calls by Benny Ricciardi and The 7 Oracles · DAEPA model
Thursday edge sheet: all 16 Week 4 games, our number vs Kalshi. Free.
No spam. Unsubscribe anytime.
- Steelers at Browns — Final: Steelers 24, Browns 27 · model had Steelers (59%, pre-game)
- Colts at Commanders — Final: Colts 30, Commanders 13 · model had Colts (65%, pre-game)
- Titans at Ravens — Final: Titans 18, Ravens 24 · model had Ravens (85%, pre-game)
- Patriots at Bills — Final: Patriots 29, Bills 26 · model had Bills (73%, pre-game)
- Jets at Bears — Final: Jets 12, Bears 23 · model had Bears (62%, pre-game)
- Jaguars at Bengals — Final: Jaguars 22, Bengals 17 · model had Bengals (59%, pre-game)
- Cowboys at Texans — Final: Cowboys 34, Texans 30 · model had Texans (60%, pre-game)
- Cardinals at Giants — Final: Cardinals 24, Giants 36 · model had Cardinals (59%, pre-game)
- Rams at Eagles — Final: Rams 24, Eagles 20 · model had Rams (62%, pre-game)
- Packers at Buccaneers — Final: Packers 17, Buccaneers 14 · model had Packers (60%, pre-game)
- Dolphins at Vikings — Final: Dolphins 10, Vikings 15 · model had Vikings (80%, pre-game)
- Chiefs at Raiders — Final: Chiefs 30, Raiders 27 · model had Chiefs (65%, pre-game)
- Chargers at Seahawks — Final: Chargers 23, Seahawks 30 · model had Seahawks (73%, pre-game)
- Broncos at 49ers — Final: Broncos 14, 49ers 24 · model had 49ers (60%, pre-game)
- Lions at Panthers — Final: Lions 26, Panthers 32 · model had Lions (62%, pre-game)
- Falcons at Saints — Final: Falcons 45, Saints 24 · model had Saints (55%, pre-game)
All 16 Week 4 games with the published win probability (fair value (books) once priced), the Kalshi price, and our pick on every game. Tap any game for the full breakdown, the model’s own line and the edge.
Who's overrated this week? Experts vs. market → — the Kalshi contracts behind every price are on the Kalshi NFL markets board, and the NFL power ratings are the numbers underneath every model line on this page.
Injury report, by unit: which defenses have lost several starters from the same front seven or secondary this week — the defensive injury cluster board.
How to use these pages: what every box on a game page means →
Live: every Week 4 game where our model and Kalshi disagree enough to call. Full edge board →
Go Pro · 26% off your first year
Every NFL game priced against the live Kalshi line, every week — $150/yr
Pro is the layer around the free game edges: fair value 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 →
Week 4 primetime, live
Run Week 4 yourself
All tools →Rankings NFL power rankings (Week 8)
Week 4 players to watch
All players →One page per player: every strike Kalshi lists, the sportsbook consensus at the same threshold, the season projection and the measured range.
Watch the Week 4 breakdown
NFL Week 4 predictions — frequently asked
Who does the model pick in NFL Week 4 2026?
Week 4 is in progress: 16 of 16 games are final and graded on this page, and the model makes no call on the 0 remaining games. The full record lives on the track record.
How are these NFL Week 4 predictions made?
Every game runs through DAEPA (Defense-Adjusted Expected Points Added), the play-by-play engine behind our PWR power ratings. It produces a neutral-field margin and a home win probability for each matchup, then the pick is wherever that probability and the Kalshi price disagree. Backtested on 3,266 games from 2013 to 2024, DAEPA picks the straight-up winner 59.6% of the time.
What does the edge on each game mean?
Edge is the published win probability — the fair value (books) once a game is priced — minus the Kalshi price for the same side, in percentage points. A +8pp edge means that number has a team at 62% while the market prices its contract at 54¢. We only call a pick when the gap clears 2 points net of the book's spread; anything smaller sits inside the noise of both numbers, so the row shows no edge.
Who will win each NFL Week 4 game?
The model's straight-up winner in each of the 16 Week 4 games, with its win probability (0 still to play; a game that has kicked off is listed as its pre-game call). Steelers over Browns (59%, pre-game); Colts over Commanders (65%, pre-game); Ravens over Titans (85%, pre-game); Bills over Patriots (73%, pre-game); Bears over Jets (62%, pre-game); Bengals over Jaguars (59%, pre-game); Texans over Cowboys (60%, pre-game); Cardinals over Giants (59%, pre-game); Rams over Eagles (62%, pre-game); Packers over Buccaneers (60%, pre-game); Vikings over Dolphins (80%, pre-game); Chiefs over Raiders (65%, pre-game); Seahawks over Chargers (73%, pre-game); 49ers over Broncos (60%, pre-game); Lions over Panthers (62%, pre-game); Saints over Falcons (55%, pre-game). The winner is the side with the higher published win probability (the fair value (books) once a game is priced); an upset is a winner priced under 50¢ on Kalshi. The straight-up winner and the play are different calls — the play is the side where that number and the price disagree by 2 points net of the book's spread.
What are the NFL Week 4 picks straight up?
Week 4 picks straight up: Steelers over Browns (pre-game call), Colts over Commanders (pre-game call), Ravens over Titans (pre-game call), Bills over Patriots (pre-game call), Bears over Jets (pre-game call), Bengals over Jaguars (pre-game call), Texans over Cowboys (pre-game call), Cardinals over Giants (pre-game call), Rams over Eagles (pre-game call), Packers over Buccaneers (pre-game call), Vikings over Dolphins (pre-game call), Chiefs over Raiders (pre-game call), Seahawks over Chargers (pre-game call), 49ers over Broncos (pre-game call), Lions over Panthers (pre-game call), Saints over Falcons (pre-game call). Each pick is the side with the higher published win probability (the fair value (books) once a game is priced) — the model's straight-up call, not a position. The play on each game is the side where that number and the Kalshi price disagree by 2 points net of the book's spread, on its matchup page.
How often are the Week 4 picks updated?
Daily through kickoff. Kalshi prices are re-captured and the model reruns against them, so the Updated stamp under the title is the last time a number on this page actually moved; it is never a render clock. Power ratings refresh weekly once games are played.
Why read this pick instead of a preview from a big sports outlet?
Two things a bigger outlet does not do. The number is posted before kickoff against the live Kalshi price, so you can see where the model and the market disagree while it still matters. And the page keeps the call after the game — result, hit or miss, closing-line value — so the record is public. Every graded call is at predictionmarketspicks.com/track-record.
Win probabilities are the market-anchored fair value (books) once a game is priced; the model’s own line is the lean. Prediction markets carry risk; contracts can settle at $0. Informational only — trade responsibly.