The model's three best-graded plays for Week 3, ranked and explained — with the caveats that come with a fourteen-game board and no long-run record to lean on.
Fourteen games got graded this week and nothing topped an A. KC at MIA is that A, and it's the model's top-ranked play by grade for Week 3. The number is a total: under 44.5. The market has this game at 44.5 points, the model has it at 41.8. That's a 2.7-point gap, and it's paired with an 8.48% edge — the second-highest edge of the three plays here, but the only one carrying the board's top grade.
A 2.7-point separation on a total isn't enormous in the scheme of NFL scoring variance, but the grade is doing the talking. When the ceiling for the week is A, and this is the game that hits it, that tells you where the model's confidence actually lives this week. Everything else on the slate graded lower than this one number.
NE at JAX is a spread play, home side at -2.5, and it carries the highest edge of the three at 8.93%. The market has the home number at 2.5. The model's own number is 5.5 — a three-point difference in the model's projected margin versus what the market is currently offering. That's a meaningful gap on a spread, and it's why this one made the ranked list even with a B grade instead of an A.
Worth sitting with for a second: this is the strongest edge percentage of the whole trio, yet it grades below the KC-MIA under. Edge and grade aren't measuring the identical thing here, and the model isn't pretending they are. A B is a real signal, just not the week's best one.
LV at NO closes the list, another total, another under: 45.5. Market sits at 45.5, model projects 43.0, a 2.5-point gap — almost identical in shape to the KC-MIA number, just graded a tier lower at B with a 7.93% edge. Two unders out of the week's top three plays is notable on its own; whether that means anything about the slate as a whole isn't something this model output speaks to, and I'm not going to invent a reason it should.
Grade B here, same as Jacksonville, lowest edge of the three. If you're ranking conviction across this list, this is the one sitting at the bottom of the podium, not off it.
Three plays, ranked, out of fourteen graded games. One A, two B's. No S, no A+ anywhere on the board this week, so nobody should be reading extra conviction into these numbers that the grades themselves don't support. This is a read on what one model is projecting against current market numbers for a single week — not a demonstrated long-term edge, and not financial advice. Treat it as one input, size accordingly, and remember a B grade is still a B grade no matter how good the write-up sounds.