Week 9 Model Picks

Week 8, if you had bet the posted card: no spreads, 0-3 on parlays, -3.0 units. Kansas City lost 28-24 in Denver and killed the first ticket. Arizona beat Dallas 33-30 and killed the second. Chicago walked into Seattle and won 34-17, which killed the third. The one spread we pulled off the card, Eagles +6 at Washington, lost by 17, so the McKee refit was right to take it off. Season to date: about -12.8 units. Here is Week 9, 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 three.

Our Bets

Spreads (0 units)

None, and this one hurts. Two edges land just under the bar. Ravens +10 against Jacksonville is a 3.8-point edge: The Model has Jacksonville by 6.2 with Joe Fagnano, an undrafted rookie with a 7.4 QBR, starting for Baltimore. Patriots -3 against Green Bay is a 3.7-point edge: The Model has New England by 6.7. Last week we trimmed a 3.8 and it saved us a unit. The bar is 4.0, so both stay off the card. If you want to bet one of them anyway, the Baltimore side is where the sharp books are (more below).

Parlays (1 unit each)

• Saints ML (-162) / Rams ML (-225) / Cowboys ML (-230). Pays +235.
• Patriots ML (-166) / Giants ML (-166) / Lions ML (-355). Pays +229.
• Bengals ML (-166) / Texans ML (-218) / Bears ML (-298). Pays +212.

The flag on the parlays is Dallas. The Model has the Cowboys by only 2.7 in Indianapolis against a 5.5-point line, and the sharp money is on the Colts side, so that is the leg we trust least. Jacksonville (-500), Kansas City (-575), San Francisco (-520), and Seattle (-455) are too short to be parlay legs.

No Thursday spread, by a tenth of a point. Jacksonville at Baltimore opened with the Ravens favored and moved about 12 points to Jaguars -10 after Lamar Jackson went on injured reserve and Tyler Huntley was lost for the season. The public is 70 to 80 percent on Jacksonville. Circa and the other sharp books are balanced or leaning Baltimore, and there is reported middle interest at Jaguars -8.5 and Ravens +10.5. The Model, with Fagnano in the file, says Jacksonville by 6.2. That jibes with the sharp side. The Under is popular at 39.5, and The Model’s total is 41.3, so it does not see that one.

The Model

Same ensemble: Ridge, LightGBM, XGBoost, and a small neural net, refit on games through Week 8, using team stats, EPA, and starter QBR. Week 8 went 10-4 straight up and 9-5 against the spread, its best spread week of the year. On the season it is 59-60-2 against the spread, and it is still behind the market on margin error, 10.0 to 9.7.

The table is the full 15-game card: per-model margin, pick, DraftKings spread, raw edge, straight-up win probability, and moneyline. Pittsburgh and Tennessee are on bye. The Model and the market pick the same winner in 14 of 15 games. The one split is Denver at Carolina, where the market has the Broncos by 2.5 and The Model calls it a coin flip, Carolina by a tenth of a point.

Quarterback notes, because this week has a lot of them. We checked every starter against this week’s reporting before running the numbers. Baltimore is Joe Fagnano. Philadelphia is Joe Milton III in his first start, with Jalen Hurts on injured reserve and Tanner McKee benched. The Giants are J.J. McCarthy with Jaxson Dart on injured reserve. Seattle is Drew Lock again, with Sam Darnold out with a quad injury and a cracked rib. Cleveland is Shedeur Sanders with Deshaun Watson in concussion protocol, and Sanders has no QBR history, so The Model is using a default for him. The one row our data source had wrong was Minnesota: it listed Kyler Murray, who is out with a hamstring, so we put Carson Wentz in and refit. That moved Buffalo at Minnesota from Bills by 2.0 to Bills by 1.6. Nothing else on the card changed.

Over/Under Picks

Totals went 5-9 in Week 8 and are 50-71 on the season. We are showing the numbers and not betting them. Three stand out, all by more than six points. Dallas at Indianapolis is 51.5 and The Model says 45.2. Tampa Bay at Chicago is 47.5 and The Model says 41.3. Arizona at Seattle is 39.5 and The Model says 46.0, an Over, which is the opposite of where the sharp money is: reports have the Under at Seattle drawing more than 80 percent of the money. The Model does not know Darnold is out when it builds the total, so trust the market on that one. Still not bets.

Where we disagree with Vegas

• JAX at BAL: JAX -10; model MOV -6.2 Jaguars. Off the card at 3.8. Covered above. The line has moved a dozen points on quarterback news, the public is piled on Jacksonville, and the sharp books are not. The Model agrees with the sharp books.
• GB at NE: NE -3; model MOV +6.7 Patriots. Off the card at 3.7. The line opened around Patriots -1.5 and moved to -3, and one report had 65 percent of the money on Green Bay getting the points. Green Bay has covered five straight. The Model is on the other side of that money: it likes New England’s roster and quarterback a lot more than the market does. Other public models reportedly make this Patriots -3 to -4, so The Model is the outlier here.
• DAL at IND: DAL -5.5; model MOV -2.7 Cowboys. Off the card at 2.8. The line opened near Cowboys -2 and steamed to -6 on 83 to 90 percent public money, with some reverse movement and Circa money showing up on the Colts. Dallas has several starters in concussion protocol. The Model agrees that the steam went too far, and it has the Cowboys winning only 58 percent of the time.
• NYJ at KC: KC -10; model MOV +7.3 Chiefs. Off the card at 2.7. We could not find much chatter on this one. The Model has Kansas City winning 71 percent of the time against a market closer to 84, so it thinks ten is rich.
• DEN at CAR: DEN -2.5; model MOV +0.1 Panthers. Off the card at 2.6. This is the only side split on the board. Public is about 73 percent on Denver and the trackers flag reverse line movement and sharp money on Carolina plus the points, with both Denver starting safeties questionable. The Model calls it a coin flip, which jibes with the sharps and not the public.
• HOU at LAC: HOU -4.5; model MOV -2.5 Texans. Off the card at 2.0. The line swung about seven points from Chargers -2 to Texans -4.5, with 90 percent of bets on Houston. The Chargers are 0-7 and depleted on the line and in the secondary. The Model still thinks 4.5 is a touch much, mostly because it does not rate Houston’s offense as highly as the market does.

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 now. Two spreads missed the card by a few tenths of a point. The rule is the rule, and last week the rule was right.

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

Good luck everyone.