Week 7 Model Picks

Week 6, if you had bet the posted card: 1-3 on spreads, 0-3 on parlays, -6.05 units. Arizona +12.5 cashed the easy way, winning outright 18-16 in Los Angeles. Everything else lost. The 2-unit Giants bet went down 34-27 to New Orleans, Jacksonville lost 34-31 to Houston in London, and New England beat the Jets 26-19 — seven points, half a point short of -7.5. All three parlays died: Tampa Bay killed the first, the Giants the second, and Jacksonville and Seattle both killed the third. Season to date: about -8.6 units. Here is Week 7, 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)

• 2 units on Giants +10 (-108). The Model has Houston by 3.3 against a 10-point line. That is a 6.7-point edge, the only number on the board that clears 4.0, and it comes with the biggest caveat we have printed this season. New York’s quarterback row is J.J. McCarthy at 77.6, and that number is one relief appearance: Jaxson Dart is out for the season after knee surgery, Jameis Winston was benched for McCarthy in the second half against the Saints, and McCarthy is reportedly the Week 7 starter, though Jim Harbaugh had not confirmed it when we ran this. If McCarthy is really a 77 passer, the Giants are the right side and ten is far too many. If that number is a half of football, the real edge is smaller than 6.7. The Giants are also banged up — Theo Johnson, Darnell Mooney, Tremaine Edmunds, Tyler Nubin, and Greg Newsome II are all on the report, and Andrew Thomas left last week with a groin issue. Houston gets Tank Dell back from injured reserve, David Montgomery is questionable, and Jalen Pitre is doubtful. The bet is “ten is too many,” not a Giants endorsement; The Model still has Houston winning.

Parlays (1 unit each)

• Browns ML (-130) / Panthers ML (-175) / Broncos ML (-230). Pays +299.
• Ravens ML (-125) / Rams ML (-218) / Jets ML (-225). Pays +279.
• Seahawks ML (-148) / 49ers ML (-175) / Vikings ML (-290). Pays +254.

Flags on the parlays. The Model still has Drew Lock starting for Seattle at 84.5 — Sam Darnold is the starter, same problem as last week, and the Seattle leg is the one we trust least. The Baltimore leg is built on Tyler Huntley’s 77.4; Lamar Jackson is day-to-day with an ankle injury and did not practice Wednesday, so Huntley is probably the right name, but if Jackson plays the number moves. Tampa Bay’s row is Jalon Daniels, and Baker Mayfield (thumb) may be back, which The Model does not know. Houston at -520 is too short to be a parlay leg.

No Thursday spread. New England at Chicago is Bears -3 and The Model calls it Chicago by 0.6, a 2.4-point edge on the Patriots that does not clear the line. Caleb Williams is back under center for the Bears. New England will probably be without Ben Brown and Alijah Vera-Tucker up front on the short week, and Drake Maye practiced in full. No edge, no bet. Baltimore -1.5 at a 3.4 edge and Philadelphia +4.5 at 2.9 stay off the card too — more on both below.

The Model

Same ensemble: Ridge, LightGBM, XGBoost, and a small neural net, refit on games through Week 6, using team stats, EPA, and starter QBR. Week 6 was a decent week for winners and a bad one for numbers: 9-5 straight up, 6-8 against the spread. Through six weeks it is 54-39 straight up and 44-48-1 against the spread. It is still behind the market on margin error, 9.8 to 9.4.

The table is the full 14-game card: per-model margin, pick, DraftKings spread, raw edge, straight-up win probability, and moneyline. Four teams are on bye — Buffalo, the Chargers, Jacksonville, and Washington. One thing worth noticing: The Model and the market pick the same side in all 14 games this week. There is not a single outright disagreement, only arguments about the size of the number.

Over/Under Picks

Totals went 7-7 in Week 6 and are 38-55 on the season. We are showing the numbers and not betting them. The biggest total edge is San Francisco at Atlanta Under 50.5. The Model says 46.5. Mike Evans is nursing an Achilles for the 49ers, which is the direction the Under needs. Still not a bet.

Where we disagree with Vegas

• NYG at HOU: HOU -10; model MOV +3.3 Texans. On the card for 2 units. The Model agrees Houston wins; it just thinks the margin is a field goal, not ten. The McCarthy number is doing a lot of the work. See above.
• CIN at BAL: BAL -1.5; model MOV +4.9 Ravens. Off the card at 3.4, and it is the week’s near miss. The Model’s Baltimore number is Tyler Huntley’s 77.4 against Joe Burrow’s 68.1. Jackson is day-to-day. Cincinnati is off a bye.
• DAL at PHI: DAL -4.5; model MOV -1.6 Cowboys. Off the card at 2.9, and the file is stale: it still has Jalen Hurts at 54.3. Hurts is on injured reserve with blood clots and Tanner McKee starts Monday night, with Joe Milton III behind him. Treat the Eagles number as a guess until McKee has a few games in the file.

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 nflverse listed. This week that bites in three places: Seattle (Lock for Darnold), Philadelphia (Hurts for McKee), and New York (a McCarthy number built on one half). Daniel Jones is at 39.4 for Indianapolis and Cam Ward is at 39.8 for Tennessee, which is most of why those two numbers look the way they do.

These are our personal bets and opinions, not betting advice. Wager at your own risk and within your means.

Good luck everyone.