Updated Sep 7, 2026 · Week 1 board.
Formula one: spread and total edge
I ran this against all 246 of our published spread and total picks and it comes back exact every time. Edge is our projected number minus the posted line. Cleveland at Jacksonville, spread market, published Aug 21, 2026. The line sat at 7.5 for the home side and we projected 8.44, so the edge is 8.44 minus 7.5. That's 0.94 points, published at 62.1% confidence (how sure the model says it is). Same game, same day, the total. Line 40.5, our projection 43.62, edge 3.12 points at 68.2% confidence. Both of those match what we published, to the decimal. That's the whole formula: subtraction you can do yourself with the two numbers already in front of you.
A second worked example, so the formula isn't a one-off
The Giants hosting Dallas, spread market. The line had the Giants at 2.5 and we projected them by 1.47, so the edge is 1.47 minus negative 2.5. That's 3.97 points at 83% confidence. Total on the same game: line 48.5, our projection 53, edge 4.5 points at 73.7% confidence. Different game, identical subtraction, identical answer every time I checked it. You don't need the model to rebuild that column. You need our number and the line, and that is the entire ingredient list. All 246 published picks reproduce the same way, not only these two.
A caveat on that Cleveland-Jacksonville line
One thing about that 7.5 before you go quoting it anywhere. It's what we had on Aug 21, 2026, and the board says something else now. Pinnacle's own number on that game went from 7.5 to 9.5 between a check on Aug 28 and today's check on Sep 7. A 2-point drift, all of it arriving after the pick published. The formula didn't move. The line did. Re-run this edge today and you use today's 9.5.
Formula two: moneyline edge, and what it skips
All 106 of our published moneyline picks come down to one line of arithmetic. Our projected win chance minus a flat 50%. Patriots at Seahawks, home side, published Jul 7, 2026. We had New England winning 59% of the time. The posted price was −185. So the edge is 0.59 minus 0.50, which is 0.09. Now notice what never showed up in that sentence. The price. Our formula treats every moneyline in the sport as a coin flip before it starts, and the −185 sitting right there never enters the arithmetic at all.
What the missing step actually costs
The version you would find in a textbook takes the vig out of the market price before anything else happens. Pinnacle had that game at home −179 and away +156, which imply win chances of 64.16% and 39.06%. Those add to 103.22%. The extra 3.22% is the piece the book keeps for itself. Scale both down so they land on exactly 100% and Pinnacle's true price on New England comes out at 62.16%. Put that next to our own 59%. The de-vigged edge is 0.59 minus 0.6216, or −0.0316. Small. Negative. Take the vig out and the market likes Seattle a hair more than our model does, which is the opposite sign from the +0.09 we printed. Both numbers are computed correctly from their own definitions. Only one of them has bothered to look at the price you'd have to pay.
| Method | Input | Result |
|---|---|---|
| Flat 50% baseline | 0.59 projected minus 0.50 flat | +0.09 |
| No-vig edge | 0.59 projected minus 0.6216 no-vig Pinnacle price | −0.0316 |
Why closing line value is the real check on either formula
Neither formula has been graded, because the opener hasn't been played. Until there are results, the way to sanity-check an edge is closing line value, meaning whether your price beat the one the market finally settled on. That check doesn't ask you to trust either formula up there. It asks you to trust the line, which moved on its own with real money behind it, and which will tell you the truth about your price long before a scoreboard does.
Do this arithmetic yourself in our EV tool. The background reading is our piece on vig and hold, then the one on closing line value. If you end up betting one of these, size it with the Kelly criterion. Tonight's board is over at our main odds screen.
The Receipts Drawer
The honest number on Patriots-Seahawks is −0.0316, and our moneyline column says +0.09. Round those off and it's a sign flip, plus 0.09 to minus 0.03, and the whole gap is a vig nobody stripped out. The spread and total side of the house holds up clean: projected minus line, 246 for 246. What would change my read on any of it: results. Nothing here has been graded, the opener hasn't kicked off, and a formula that has never been wrong has also never been tested. So track closing line value as the first games post. That's the number that tells you which of these two is worth your money on the next slate.
