Prices as of Tuesday, September 8, 2026, the board every model read before any Week 1 game kicked off; the Bet now links open the book's live number, which may have moved.
As of Friday, September 11, 2026, 7 AI models have each picked three NFL Week 1 player props from the same written playbook and the same data packet: every posted line at every book, usage, coverage, personnel changes, coaching history, injuries and weather for all 16 games. Nobody told them who to pick, no model saw another's picks, and every model had to explain each pick in its own words. Every pick is graded against the box score, and each model's profit is tracked in units all season. This is Week 1; the standings start at zero.
Stay ahead of the markets.
Daily insights and expert picks on Kalshi, Polymarket, and what's moving markets.
Free forever. Unsubscribe anytime.
The Read
Seven models read the same sixteen-game board and came back with 21 plays between them, and they did not agree on much: two landed on the same Patriots running back's workload from opposite ends of the reasoning, one built its three around names most bettors skip, and one took the quarterback under nobody else touched. The picks are below in each model's own words, with a link to the bet. The record that matters starts Monday.
How it works
Every model gets two things: Dave Loughran's playbook for researching props, which says what to look at, what counts and what to ignore, and a data packet covering the whole slate. Each model picks three, writes its reasoning for each, and hands back the list. How we keep score: every bet is to win one unit. At -130 that means risking 1.3 units to win 1. At +130 it means risking 1 unit to win 1.3. If your unit is $100, -130 is $130 to win $100, and +130 is $100 to win $130. Wins and losses add up week to week.
The Standings
| Model | Record | Units | Weeks |
|---|---|---|---|
| Claude Opus 5 | 1-0 | +1.00 | 1 |
| Claude Fable 5.1 | 0-0 | +0.00 | 1 |
| OpenAI GPT (codex) | 0-0 | +0.00 | 1 |
| Google Gemini Pro | 0-0 | +0.00 | 1 |
| Moonshot Kimi | 0-0 | +0.00 | 1 |
| Claude Sonnet 5 | 0-0 | +0.00 | 1 |
| xAI Grok | 0-1 | -1.00 | 1 |
Ranked by units won. Every bet is to win one unit.
The Lines
| Model | Pick | Price | Book |
|---|---|---|---|
| Claude Opus 5 | Rhamondre Stevenson under 60.5 rushing yards | -115 | BetRivers |
| Claude Opus 5 | Jacoby Brissett over 34.5 pass attempts | 104 | Caesars |
| Claude Opus 5 | Colston Loveland over 50.5 receiving yards | -114 | FanDuel |
| Claude Fable 5.1 | Kyren Williams over 53.5 rushing yards | -110 | DraftKings |
| Claude Fable 5.1 | Jaylen Warren over 40.5 rushing yards | -114 | FanDuel |
| Claude Fable 5.1 | Mark Andrews under 36.5 receiving yards | -115 | BetRivers |
| OpenAI GPT (codex) | Malik Willis under 185.5 passing yards | -116 | WynnBET |
| OpenAI GPT (codex) | Darius Slayton over 14.5 receiving yards | -110 | Bet365 |
| OpenAI GPT (codex) | Kenny Gainwell over 20.5 receiving yards | -114 | FanDuel |
| Google Gemini Pro | Kyle Pitts over 39.5 receiving yards | -110 | Bet365 |
| Google Gemini Pro | David Montgomery over 12.5 rush attempts | -130 | Unibet |
| Google Gemini Pro | Patrick Mahomes under 225.5 passing yards | -110 | Fanatics |
| Moonshot Kimi | Christian McCaffrey over 39.5 receiving yards | -105 | PointsBet |
| Moonshot Kimi | Javonte Williams over 73.5 rushing yards | -113 | Caesars |
| Moonshot Kimi | Kenneth Walker III under 16.5 rush attempts | -115 | Fanatics |
| Claude Sonnet 5 | Courtland Sutton over 39.5 receiving yards | -130 | Bet365 |
| Claude Sonnet 5 | Derrick Henry over 79.5 rushing yards | -102 | Caesars |
| Claude Sonnet 5 | Xavier Worthy under 37.5 receiving yards | -114 | DraftKings |
| xAI Grok | Rhamondre Stevenson under 15.5 rush attempts | 100 | DraftKings |
| xAI Grok | Cam Skattebo over 14.5 receiving yards | -110 | Bet365 |
| xAI Grok | J.K. Dobbins over 49.5 rushing yards | -105 | Bet365 |
Claude Opus 5
1. Rhamondre Stevenson under 60.5 rushing yards, -115 on BetRivers. Bet now at BetRivers
Seattle's run defense is the most extreme unit on the board, and it's extreme on the three metrics that decide a rushing-yards line rather than one. The Seahawks finished last in the league in rush EPA per play allowed (-0.205, #32/32), last in yards per carry (3.40, #32/32) and last in explosive run rate allowed (4.8%, #32/32), with a 22.3% stuff rate (#6) and 74.8 rushing yards per game allowed to running backs. They did it playing the lightest boxes in football — an average box of 6.28 (#31), a 74.0% light-box rate — which is the tell that this is personnel and technique, not eight men at the line.
2. Jacoby Brissett over 34.5 pass attempts, +104 on Caesars. Bet now at Caesars
The cleanest volume market on the board, and the price is half the edge: Caesars is +104 on a number Pinnacle prices at -102. Arizona led the league in pass rate over expectation on neutral early downs at 7.0, threw on 61.2% of neutral snaps (#2) and averaged 46.5 dropbacks a game. Brissett averaged 34.6 attempts and went over 34.5 in 9 of 14 starts, averaging 34.8 over his last four while playing 100% of the snaps in that stretch.
3. Colston Loveland over 50.5 receiving yards, -114 on FanDuel. Bet now at FanDuel
This is the deepest case of any receiving spot on the board, and the role change is only half-priced. Chicago vacated 28% of its 2025 targets — DJ Moore's 85 and Olamide Zaccheaus's 65 — and added nobody to claim them: Kalif Raymond arrives as WR3 off 30 targets in Detroit, and the only other additions at the position are rookies. That share goes to Odunze, Burden and Loveland, whose snap share was already climbing, from 64.5% on the season to 80.2% over the last four games, where he averaged 69.5 receiving yards.
Claude Fable 5.1
1. Kyren Williams over 53.5 rushing yards, -110 on DraftKings. Bet now at DraftKings
Kyren Williams cleared 53.5 rushing yards in 15 of 17 games in 2025, averaged 73.6, and his season range was 46–114 — a floor two yards under this number, and he's 4 of 4 over it in his last four (75.0 average). That production came behind a wrecked line: the Rams return four starters who barely played — Steve Avila (1 of 17 games), Kevin Dotson (2), Coleman Shelton (3), Alaric Jackson (3) — so 73.6 per game is closer to a floor than an artifact. The scheme rate that survives Robert Saleh's return is the one that helps: San Francisco's average box of 6.09 was the league's lightest and its 83.4% light-box rate the highest, and Williams took 56.4% of his carries against light boxes.
2. Jaylen Warren over 40.5 rushing yards, -114 on FanDuel. Bet now at FanDuel
The line is 19 yards below Warren's own 2025 average. He cleared 40.5 in 11 of 16 games (69%), averaged 59.9 rushing yards on a 55.4% carry share, and closed the year at 76.5 per game over his last four with 3 of 4 overs. The discount is the Rico Dowdle arrival, but it's mispriced: Pittsburgh vacated 37% of its 2025 carries (Kenny Gainwell's 114 and Kaleb Johnson's 28 both departed), which is the share Dowdle, listed RB2, inherits — Warren's lead role isn't what left the room.
3. Mark Andrews under 36.5 receiving yards, -115 on BetRivers. Bet now at BetRivers Shown at BetRivers at the same number; the model named Unibet, which most readers cannot access.
The biggest gap between a posted number and a player's own distribution of the three plays here. Andrews averaged 24.8 receiving yards in 2025 and cleared 34.5 twice in 17 games (12%); he went 0 for 4 over it down the stretch at 20.2 per game, on a snap share that fell from 61.7% to 46.2% in the last four. The line sits ten-plus yards above his average because the market is paying him Isaiah Likely's vacated 9% target share — but Baltimore spent pick 80 on Ja'Kobi Lane, now the WR3, so the room added a pass catcher rather than clearing one out.
OpenAI GPT (codex)
1. Malik Willis under 185.5 passing yards, -116 on WynnBET. Not graded: this market was not on the board the models were given.
The player profile, team structure, personnel change and matchup all point the same way for Willis. He averaged 105.5 passing yards in 2025, went over 174.5 only 1 time in 4 games and averaged just 8.8 pass attempts. Miami’s offense was already slow and run-leaning: 37.8 seconds per snap, 56.8 plays per game, 32.8 dropbacks per game, 49.9% neutral pass rate and -3.6 PROE.
2. Darius Slayton over 14.5 receiving yards, -110 on Bet365.
Slayton is the best low-number receiving over because he does not need a major target share to clear. His line is 14.5 yards at Bet365, while his 2025 average was 38.4, and he cleared this number in 12 of 14 games. Dallas is the matchup that makes the number look most wrong: the Cowboys allowed 271.2 pass yards per game, 171.0 receiving yards per game to wide receivers, 9.63 yards per target to wide receivers, 8.32 YPA and an 8.8% explosive pass rate.
3. Kenny Gainwell over 20.5 receiving yards, -114 on FanDuel. Bet now at FanDuel
The role, team need and defensive leak all point at the same specific market for Gainwell. He had 85 targets, 73 catches and 486 receiving yards with Pittsburgh in 2025, clearing 20.5 receiving yards in 9 of 17 games with a 28.6-yard average. His late-season role was stronger: 60.2% snaps over his last four and 50.0 receiving yards per game.
Google Gemini Pro
1. Kyle Pitts over 39.5 receiving yards, -110 on Bet365. Bet now at Bet365
The arrival of offensive coordinator Tommy Rees completely changes the baseline opportunity for Atlanta's tight ends. During his 2025 stop in Cleveland under Kevin Stefanski, Rees helped direct a scheme that ranked second in the league in both tight end target share (32.4%) and tight end targets per game (10.0), while funneling 36.8% of red-zone looks to the position. Pitts already commands an elite 22.7% target share and 0.220 targets per route run, making him the perfect focal point for this system upgrade.
2. David Montgomery over 12.5 rush attempts, -130 on Unibet. Not graded: this market was not on the board the models were given.
A run funnel mechanism dictates rushing volume before efficiency, and the Buffalo defense is the most extreme run funnel on the board. The Bills allowed the second-most expected points added (EPA) per rush (0.059) and the third-most yards per carry (4.92), while seeing the fifth-highest pass rate faced (57.3%), meaning opposing offenses abandon the pass to attack them on the ground. Houston added David Montgomery to inherit the early-down role after Nick Chubb vacated 26% of the team's carries, while Buffalo's defensive front lost linebacker Matt Milano, meaning their already leaky run defense should be even softer up the middle.
3. Patrick Mahomes under 225.5 passing yards, -110 on Fanatics. Bet now at Fanatics
Volume drives passing props, and while the Chiefs led the league with a 62.2% neutral pass rate, this number asks Patrick Mahomes to clear an average bar against an absolute outlier of a pass defense. The Broncos suffocated opposing quarterbacks last season, allowing an NFL-low 5.83 yards per attempt and a league-bottom 5.6% explosive pass rate. They generate pressure at an elite clip, ranking first in both sack rate (9.1%) and quarterback-hit rate (20.0%).
Moonshot Kimi
1. Christian McCaffrey over 39.5 receiving yards, -105 on PointsBet. Bet now at PointsBet
This is where the most independent factors line up in the same direction. The distribution outlier: McCaffrey cleared this number in 13 of 17 games last year and averaged 54.4 receiving yards — the line sits 15 yards under his average, and Pinnacle agrees it's cheap, hanging 38.5 while the market sits at 39.5. The volume claim is structural, not a hot streak: 129 targets, a 23.3% target share, 7.6 targets per game, and Kyle Shanahan is still calling it — his 2025 offense directed 30.5% of its targets at running backs, the highest rate in the league.
2. Javonte Williams over 73.5 rushing yards, -113 on Caesars. Bet now at Caesars Shown at Caesars at the same number; the model named Unibet, which most readers cannot access.
The outlier here is the metric that decides this market, and it is as extreme as the board offers this week. The 2025 Giants finished dead last in the league in rushing EPA allowed (0.127, #1/32), yards per carry allowed (5.30, #1/32) and explosive run rate allowed (15.6%, #1/32), with 129.9 rushing yards per game (#2/32). The roster losses point the same way: that league-worst unit was built WITH Dexter Lawrence (17 games, 67% of snaps) and Bobby Okereke (99% of snaps), and both are gone, along with Cordale Flott and Dane Belton.
3. Kenneth Walker III under 16.5 rush attempts, -115 on Fanatics. Bet now at Fanatics
This under stacks on the volume side. Kansas City was the most pass-heavy offense in football last year: a 62.2% neutral pass rate (#1/32), PROE of 5.9 (#3/32), only 21.4 rush attempts per game for the entire team. Walker's own distribution is the second outlier: as Seattle's lead back in 2025 he topped 16.5 carries twice in 17 games (12%), averaging 13.0 on a 46.9% carry share in an offense that ran it 28.6 times a game — five more team carries per game than the one he just joined.
Claude Sonnet 5
1. Courtland Sutton over 39.5 receiving yards, -130 on Bet365. Bet now at Bet365
Sutton cleared this number in 12 of 17 games last season (71%) and in 3 of his last 4 (75%, averaging 61.0) — and the line here is 39.5, more than 20 yards below his season average of 59.8. He's Denver's clear top target (21.2% target share, 7.4 targets/game) and a red-zone fixture (19.0% of RZ targets, 35.9% of end-zone targets). Kansas City's pass defense already allowed 7.34 yards per target to wide receivers (#27) in 2025, and has since lost three defensive backs who combined for the bulk of last year's secondary snaps — Trent McDuffie (90% of snaps), Jaylen Watson (95%) and Bryan Cook (82%) are all gone.
2. Derrick Henry over 79.5 rushing yards, -102 on Caesars. Bet now at Caesars
Henry carries 68.1% of Baltimore's rush load, 63.0% of its red-zone carries and 80.0% of its inside-the-10 carries on the league's most run-heavy offense (neutral-script pass rate 47.3%, #31; PROE -8.2, #31). Indianapolis's run defense allowed 3.76 yards per carry (#27) with a front seven that no longer exists — Zaire Franklin (95% of defensive snaps) and Kwity Paye (62%), the Colts' two most-used front-seven defenders, both departed this offseason. Henry cleared 79.5 in 9 of 17 games (53%), averaging 93.8 yards on the year.
3. Xavier Worthy under 37.5 receiving yards, -114 on DraftKings. Bet now at DraftKings
Denver allowed the lowest yards-per-play mark in the league in 2025 (4.24, #32), the lowest explosive-pass rate (5.6%, #32) and the second-stingiest deep-completion rate (28.0%) — and unlike several units on this board, this isn't an injury mirage: core pieces like Nik Bonitto, Alex Singleton and Justin Strnad each played 15-plus games and return intact. Worthy's whole profile is vertical (12.9 aDOT, 17 targets of 20-plus air yards), and he still cleared this number in only 7 of 14 games (50%) even against an average slate, with his last four games averaging 32.8 yards.
xAI Grok
1. Rhamondre Stevenson under 15.5 rush attempts, +100 on DraftKings. Bet now at DraftKings
Stevenson carried 9.4 times per game in 2025 (carry share 32.8%) and cleared 15.5 attempts in 1 of 14 games; the over is a 7% proposition against his own distribution. TreVeyon Henderson already owned 45.0% of New England's carries (10.6/g), so the room is a committee before anyone lines up in Seattle. Macdonald's Seahawks allowed the fewest rush yards per carry in the league (3.40, #32), the worst rush EPA per play (#32), and the lowest explosive-run rate (#32), with a 22.3% stuff rate — scheme rates under the same coordinator, not a one-year fluke.
2. Cam Skattebo over 14.5 receiving yards, -110 on Bet365. Bet now at Bet365
Skattebo averaged 25.9 receiving yards in eight games (4.0 targets per game) and 27.2 over his last four, yet the lowest over on the board is 14.5. Dallas funneled work to running backs through the air — 5.9 targets, 5.0 receptions and 43.2 receiving yards per game to RBs (#3) at 7.28 yards per target (#3) — and played zone on 72.2% of snaps. Skattebo's own numbers back it up: against zone he ran 2.06 yards per route run on 90 routes with a 25.6% target rate (23 targets, 20 catches, 185 yards); against man he was ordinary (0.6 Y/RR).
3. J.K. Dobbins over 49.5 rushing yards, -105 on Bet365. Bet now at Bet365
Dobbins cleared 49.5 in 9 of 10 games in 2025 at 77.2 yards per game, and the last four sat at 82.5 on 15.5 carries. That is a raw number-versus-line gap you do not invent — the board is still hanging a 49.5. Opportunity is the early-down job: 15.3 carries per game, 38.0% carry share, 7.4 neutral-script carries per game when he was active.
Bottom Line
7 models, 21 plays, one board. The standings above are the point of the series and they start at zero; the picks are the models' own, and the full write-up behind every one is on file. Nothing here is a Dave Loughran pick unless it also appears on one of his pages.
Follow Dave on X: @Loughy.
Hero illustration: OddsShopper, in the house collage style. Derrick Henry photo by Tennessee Titans, licensed CC BY 3.0; Christian McCaffrey photo by Tennessee Titans, licensed CC BY 3.0; photos cropped, toned, and composited.
Odds quoted are from the OddsShopper live-odds board as of September 8, 2026, and will move before kickoff; always check the number at your book before betting. OddsShopper is an independent analyst: the picks above are AI model picks, and any figures referenced are model estimates, not predictions of fact and not financial advice. Must be 21+ to wager at a sportsbook (18+ on prediction markets); availability varies by state. Betting involves risk. If you or someone you know has a gambling problem, call 1-800-GAMBLER.
