The prices are live.The model is not.
Every moneyline, spread and total on the Week 1 board, de-vigged book by book into a consensus fair number, then inverted against your +6% EV gate so each market shows exactly what a model would have to believe before it becomes a play. There is no football analogue yet to the starter projections and park factors that drive the baseball model, so the model column reads Pending on every row and Week 1 is a no-action week. What is real today: the price screen, the hold, the line-shopping value, and the frozen opening numbers that make closing-line value measurable from here on.
Every market, priced against the gate
Fair % is the average no-vig probability across the books quoting that exact number — never a blend across two different spreads or totals. Needed is the probability your model would have to give the selection at the best available price to clear +6%, rounded against the bettor. Need price flips it the other way: the price the market would have to offer for the fair number itself to clear. Drop a projection next to either column and the row decides itself.
| Kickoff | Game | Market | Selection | Best | Book | Fair % | Fair | Model % | Needed | Need price | Books | Status |
|---|
Books is how many sportsbooks had two sides on that number and therefore fed the fair price. A market quoted by one book is still shown, but its fair number carries that book's own lean and is not a consensus.
What the same bet pays at different windows
Identical selection, identical number, different book. This is the one edge on the page that needs no model at all — it is arithmetic on posted prices. Gain is the extra return on the same stake from taking the best price instead of the worst.
| Game | Market | Selection | Best | Book | Worst | Book | Gain |
|---|
The full Week 1 board
| Kickoff | Game | Venue | Spread | Total | Favourite | Fair win % | Best ML | Hold |
|---|
Favourite and fair win % are read straight off the de-vigged moneyline — the market's opinion, not the page's. Hold is the average vig across the three books on that game's three markets.
Where the books disagree on the line itself
Football books do not all hang the same spread or total. A book sitting on a different number is pricing a different bet, so it is excluded from that market's fair probability rather than averaged into it — but it is listed here, because a half-point is often worth more than the price.
| Game | Market | Consensus | Off-line book | Its number | Its quotes |
|---|
Passing, rushing, receiving and receptions
Every prop below is context only. None of it is tracked, none of it is graded and none of it can be promoted to an official play — the same rule the baseball page applies to its home-run and strikeout boards. What these rows do is show you where the price is and what the three books collectively think, so you can see the number rather than take someone's word for it.
| Player | Game | Market | Line | Lean side | Best price | Book | Fair % | Needs | Edge | Hold | Gate |
|---|
The market with only one side
| Player | Game | Best price | Book | Raw implied | Shop gain | Books posting | Gate |
|---|
Raw implied still has the book's margin inside it, so it overstates the real chance — that is exactly why no fair number is printed here. Sorted by shortest price, which is the market's own ranking of who is most likely to score.
Frozen for closing-line value
Every rebuild writes the consensus spread, total and favourite win probability to a per-game history file. A weekly sport gives the number days to move, which makes closing-line value worth measuring — the single most honest test of whether a model is any good.
| Game | Open spread | Now | Open total | Now | Favourite | Open win % | Now | Move |
|---|
Every tracked selection, graded
| Week | Game | Market | Selection | Price | Book | EV at entry | Result | Units |
|---|
Graded the morning after each week's games from official settlement, at a flat 1 unit per tracked selection. Settlement wording is preserved exactly — Won, Lost, Refunded, Half Won, Half Lost. A week with no qualifying play is logged as No action, not skipped, and the NFL log is kept entirely separate from the baseball one.
How this page is built
Consensus de-vig, one number at a time
Each book's two-sided market is stripped of hold with p = (1/d) / Σ(1/d), then averaged only
across books quoting the same spread or total. Two different posted numbers are two different markets and
are never averaged together.
+6% EV, rounded against you
Required probability is 1.06 / decimal price; the required price is
(1.06 / p) converted back to American. Plus-money requirements round up, minus-money
magnitudes round down — the same single rounding rule used everywhere.
The model tier is honestly empty
The baseball model runs on published event probabilities plus starter and park inputs. Football has no equivalent wired in, and the run-distribution engine does not transfer — points arrive in lumps of 3 and 7 with spikes on the key numbers, which a baseball-shaped distribution gets badly wrong. So every row is Pending and Week 1 is a no-action week. That is the correct output, not a missing feature.
What would change it: a published per-game win probability, the same way the baseball board gets one. Given that single input, the gate column on this page turns itself on with no further work — the screen, the rounding, the snapshot and the grading pass are already built.