Week 10 Model Picks
Week 9, if you had bet the posted card: no spreads, 2-1 on parlays, about +3.5 units. Finally. The Saints/Rams/Cowboys ticket cashed at +235, and New Orleans needed every point of a 23-22 win over Cleveland to get there. The Bengals/Texans/Bears ticket cashed at +212. The Patriots/Giants/Lions ticket died in New England, where Green Bay won 45-27. The two spreads we left off the card at 3.8 and 3.7, Ravens +10 and Patriots -3, both lost, so the 4.0 bar earned its keep again. Season to date: about -9.3 units. Here is Week 10, on the same rules. A spread has to clear a raw edge of 4.0. Totals stay on the page and off the card. Three 1-unit moneyline parlays. The budget is eight units. This week spends five.
Our Bets
Spreads (2 units)
- 2u Packers -3 (-105) vs. Minnesota. The Model has Green Bay by 7.3, a 4.3-point edge, and gives the Packers a 71 percent chance to win against a market near 58.
Parlays (1 unit each)
- Panthers ML (+110) / 49ers ML (-205) / Texans ML (-225). Pays +351.
- Giants ML (-122) / Bengals ML (-180) / Ravens ML (-230). Pays +306.
- Packers ML (-148) / Chiefs ML (-180) / Seahawks ML (-298). Pays +248.
The flag on the parlays is Carolina. It is the only leg where The Model picks the underdog to win outright, Panthers by 1.1 in New Orleans against a 2.5-point line, and at 53 percent it is barely a lean. The Giants leg goes first, tonight against Washington. Buffalo, Jacksonville, and the Rams are shorter than -400, so they stay out of the parlay pool.
Two spreads missed the card. Carolina +2.5 at New Orleans is a 3.6-point edge, and Miami +3.5 at Indianapolis is a 3.6-point edge. Miami is starting Kyle McCord, who has no QBR history, so The Model is using a default for him. That makes the Dolphins number softer than it looks. The bar is 4.0, so both stay off.
The Model
Same ensemble: Ridge, LightGBM, XGBoost, and a small neural net, refit on games through Week 9, using team stats, EPA, and starter QBR. Week 9 was a strange one: 13-2 straight up, its best winner-picking week of the year, and 4-11 against the spread, its worst. It had the right teams and kept underrating the margins. On the season it is 63-71-2 against the spread, and it is still behind the market on margin error, 10.0 to 9.6.
The table is the full 14-game card: per-model margin, pick, DraftKings spread, raw edge, straight-up win probability, and moneyline. Chicago, Denver, Philadelphia, and Tampa Bay are on bye. The Model and the market pick the same winner in 12 of 14 games. The splits are Carolina at New Orleans (The Model has the Panthers by 1.1) and Miami at Indianapolis, where it calls the game a dead coin flip against a 3.5-point Colts line.
Quarterback notes. We checked every starter against this week’s reporting, and our data source had four of them wrong. Seattle is Drew Lock again, with Sam Darnold still out. Miami is Kyle McCord in his first start, with Malik Willis on injured reserve. Arizona is rookie Carson Beck, with Gardner Minshew on injured reserve with a torn UCL. Baltimore is Jacoby Brissett, acquired from Arizona last week, with Lamar Jackson and Tyler Huntley both on injured reserve. We put all four in and refit. The big mover was Baltimore: swapping Huntley for Brissett took the Ravens from 6.6 to 5.9 over the Chargers and dropped that game out of the play tier. McCord and Beck have no QBR history, so they get the default. Justin Herbert’s finger X-rays were negative and he is off the injury report. Jared Goff practiced in full Thursday ahead of the Munich game.
Over/Under Picks
Totals went 7-8 in Week 9 and are 57-79 on the season. We are showing the numbers and not betting them. Three Overs stand out. Seattle at Las Vegas is 40.5 and The Model says 48.9. Washington at the Giants is 40.5 and The Model says 45.8. The Rams at Arizona is 39.5 and The Model says 44.0. The Model has leaned Over on low totals all season and it has not paid, so treat these as curiosities. Still not bets.
Where we disagree with Vegas
- MIN at GB: GB -3; model MOV +7.3 Packers. On the card at 4.3. The biggest gap on the board. The Model rates Jordan Love and the Packers’ efficiency numbers well above Minnesota’s.
- CAR at NO: NO -2.5; model MOV -1.1 Panthers. Off the card at 3.6. A side split. The quarterbacks grade out nearly even, and The Model gives New Orleans less home-field credit than the market does.
- MIA at IND: IND -3.5; model MOV -0.1 Dolphins. Off the card at 3.6. The other split, and the softer one, because McCord is a default QBR. Daniel Jones’s 37.9 QBR is doing a lot of the work here.
- PIT at CIN: CIN -3.5; model MOV -6.0 Bengals. Off the card at 2.5. The Model likes Joe Burrow’s 75.3 QBR against Aaron Rodgers’s 48.1 more than the market does.
- KC at ATL: KC -3; model MOV -5.0 Chiefs. Off the card at 2.0. Mahomes is a full go, and The Model has Kansas City winning 65 percent of the time.
Remember, the best way to use this is as one input into your own thinking. The Model does not know who is holding out, and it only knows the quarterback our data source lists, which is why we check every row by hand. This week the hand check caught four.
These are our personal bets and opinions, not betting advice. Wager at your own risk and within your means.
Good luck everyone.