NFL Week 3 picks
Week 3 is in the books. Here is the free pick as it was published, and how every pick on the members' board landed — wins and losses alike, graded at the published price.
NO ML -168 at FanDuel
Las Vegas Raiders at New Orleans Saints · Sunday, September 27 · 4:25 PM ET · final LV 35 – NO 27
The model projects New Orleans Saints by 6.8 points (18–25, LV at NO), with a projected total of 43 against the market's 43.5. The market has New Orleans Saints favored by 3.5. On the moneyline the model's raw number gives New Orleans Saints a 76% chance to win; the price of -168 implies 61.1%. We never bet the raw model — the market is part of the number. Blended, our probability is 67.2%, an edge of 6.2% over the price. That is what makes it the free pick: of the plays still to come on this week's board priced between -250 and +250 whose number beats the price even after the sportsbook's cut, this is the most likely to win.
How it landed: LOSS · LV 35 – NO 27
The line since we posted it: closed -175 · beat the close by 7¢.
Published 2026-09-22 at the price shown; the pick is kept as published and graded at that price, even if the line moves.
Get the free pick by email every week free account, no card
Last week: safest bets went 10–6 (-1.76 units), value bets 5–11, winners called 9 of 16. Week 2 pick by pick →
How Week 3 landed
| Result | Pick | Price | Our probability | Final |
|---|---|---|---|---|
| WIN | BUF ML safest | -325 | 79.7% | LAC 16 – BUF 24 |
| WIN | KC ML safest | -550 | 78.3% | KC 24 – MIA 10 |
| LOSS | SEA ML safest | -380 | 77.3% | SEA 31 – WAS 33 |
| WIN | SF ML safest | -375 | 76.5% | ARI 30 – SF 36 |
| WIN | DET ML safest | -290 | 75.1% | NYJ 24 – DET 31 |
| LOSS | NO ML safest | -175 | 66.6% | LV 35 – NO 27 |
| WIN | BAL ML safest | -162 | 64.9% | BAL 34 – DAL 31 |
| LOSS | PHI ML safest | -175 | 60.7% | PHI 7 – CHI 27 |
| LOSS | CIN ML safest | -170 | 59.5% | CIN 27 – PIT 30 |
| WIN | NYG ML safest | -130 | 57.3% | TEN 7 – NYG 12 |
| WIN | JAX ML safest | -150 | 55.6% | NE 6 – JAX 35 |
| WIN | Over 44 safest | -110 | 55.3% | LA 26 – DEN 30 |
| LOSS | Over 41.5 safest | -110 | 54.3% | CAR 18 – CLE 21 |
| LOSS | TB +1.5 safest | -113 | 54.2% | MIN 23 – TB 16 |
| LOSS | HOU ML safest | -122 | 54.1% | HOU 17 – IND 19 |
| LOSS | GB ML safest | -240 | 67.6% | ATL 35 – GB 14 |
| LOSS | MIA ML value | +500 | 21.7% | KC 24 – MIA 10 |
| WIN | Over 40.5 value | -105 | 58.8% | SEA 31 – WAS 33 |
| WIN | BUF -7 value | -105 | 56.6% | LAC 16 – BUF 24 |
| LOSS | Over 37.5 value | -105 | 55.2% | TEN 7 – NYG 12 |
| WIN | Over 44 value | -110 | 55.3% | LA 26 – DEN 30 |
| LOSS | NO -3.5 value | -102 | 53.1% | LV 35 – NO 27 |
| WIN | BAL ML value | -162 | 64.9% | BAL 34 – DAL 31 |
| LOSS | NE ML value | +136 | 44.4% | NE 6 – JAX 35 |
| WIN | CHI ML value | +164 | 39.3% | PHI 7 – CHI 27 |
| LOSS | Over 41.5 value | -110 | 54.3% | CAR 18 – CLE 21 |
| WIN | PIT ML value | +155 | 40.5% | CIN 27 – PIT 30 |
| WIN | DET -6.5 value | -110 | 53.7% | NYJ 24 – DET 31 |
| LOSS | Over 42.5 value | -106 | 52.7% | HOU 17 – IND 19 |
| LOSS | ARI ML value | +335 | 23.5% | ARI 30 – SF 36 |
| LOSS | TB +1.5 value | -113 | 54.2% | MIN 23 – TB 16 |
| WIN | Over 43 value | -110 | 53.1% | ATL 35 – GB 14 |
One unit per bet at the published price. Parlays, win + kicker combos and the parlay of the week are graded the same way on the board's Record tab.
About these picks
What is the free pick?
The most likely winner on the NFL board this week that is still priced in your favour: a moneyline between -250 and +250 where our number — the model blended with the market price — beats the sportsbook's price even after the book's cut (if no moneyline qualifies, the most likely spread or total that does). It is chosen when the week's board is first published, kept as published, and graded in public after the game like every other pick. Through Week 2 of 2026 the free pick was the single most likely side on the board instead; those are shown as they were published.
Where do the probabilities come from?
A proprietary model projects every game's margin and total from team strength, recent form, injuries and matchups, then blends its raw probability with the price the sportsbooks are offering. The blended number is what we publish; the raw model on its own would call winners a touch less often than the closing line, which is why the market is part of the number.
Does a high probability mean the bet is safe?
No. 67.2% is a probability, not a promise: the other side happens. Favorites pay short prices for exactly that reason. Black Box Bets is an information service, not a sportsbook; no pick, parlay or strategy is guaranteed to profit, and past results — including the 2018–2025 test — do not predict future ones.