What a letter
actually looks like.
This is a real issue — Week 1 of the 2026 season — with the bets taken out. Everything else is exactly what lands in your inbox the morning of the games.
Three layers, and only one of them is a bet
Every game on the slate gets its own page, built in three parts that are deliberately unequal in weight. Only the last is a play, and only the last is withheld here.
- The numbers. How both teams actually performed last season, with the league rank beside every figure — so a number means something without you having to know the league average.
- The read. What the data says about markets the letter does not publish a bet in: team totals, sacks, volume, turnovers. It states a fact with a sample and stops. It never names a side, because none of it has been graded against a price.
- The bets. The only part that names a side, and it never appears without the worst price still worth taking it at.
That separation is the whole product. A pick without a price is an opinion — the same pick is a winning bet at one number and a losing one at another.
NE at SEA
SEA −3.5 · total 44.5 · I projectNE and SEA numbers are 2025 regular season — small figure is league rank
| Metric | NE | SEA |
|---|---|---|
| EPA per play | 0.169 1 | 0.045 10 |
| EPA allowed | -0.051 11 | -0.144 1 |
| Success rate | 0.484 3 | 0.466 6 |
| Explosive play rate | 0.085 1 | 0.077 3 |
| Points per drive | 2.73 4 | 2.50 8 |
| Drive score rate | 0.456 6 | 0.466 5 |
| Seconds per play | 29.7 30 | 29.0 21 |
| Pass rate over exp. | 2.0 4 | -6.7 30 |
| Pressure rate | 0.276 24 | 0.341 4 |
| Takeaways / game | 1.1 19 | 1.4 6 |
The read — markets I don’t publish
Team totalNE scored 2.73 points per drive (4th) on 10.6 drives a game, starting on average at their own 30.
Sacks / QBSEA pressured on 34% of dropbacks (4th) and averaged 2.8 sacks a game. D.Maye went down on 8.7% of attempts throwing 9.1 yards downfield on average.
VolumeSEA's pass rate over expected was 8.4 points lower when trailing than when leading. That is measured against a baseline that already knows the score, so it is them leaning run relative to what other teams do from the same spot. They are the favorite here at 3.5.
In the letter this is where that game's plays sit, each naming the side, the number and the worst price it is still worth taking at. Most games have none — that is the model declining rather than failing.
The letter opens with the short list
Before any of the game pages: one ranked list of the best bets on the whole board that week, across every market, each with the price it needs. It is usually short. In a normal week most of the board clears nothing at all.
Ranked across all fifteen markets, never more than five, and capped so one market cannot take every slot. Every row carries a price to beat — a row without one would be an opinion, and those do not go in.
And the parts that are not about one game
Teaser legs come last, because a teaser is not a bet until it is two or three legs from different games. Each leg is published with the number to take it at, the book hanging that number when it beats the market’s, and the worst payout worth accepting on a two- or three-team ticket.
The guide on teasers explains the arithmetic behind that section in full, with the sample size on every figure. That part is not gated.
Get the whole thing, free, every game day
Every game that kicks off that day, the short list at the top, and the price beside every bet — hours before kickoff, while the number is still on the board.
The stats and the read above are reproduced verbatim from the Week 1 build. They are public NFL figures from the 2025 regular season and market lines from the board — nothing on this page is a projection or a play. 21+, and betting loses money on average.