Updated Sep 7, 2026 · Week 1 board.
The subtraction that defines edge
In our published picks, edge is that same subtraction every single time. Our projected number for the game, less the market's line at the moment we made the pick. That's the whole formula on a spread or a total. Take one away from the other and the remainder is your edge, counted in points.
It holds exactly in the numbers, which is the part I enjoy. Take the Week 1 game between the Cleveland Browns and the Jacksonville Jaguars. Our published spread pick carried a market line of 7.5, our own projection of 8.44, and a stored edge of 0.94. Check it: 8.44 − 7.5 = 0.94. The total pick on the same game carried a market line of 40.5, our projection of 43.62, and a stored edge of 3.12. Check that one too: 43.62 − 40.5 = 3.12. Nothing there is estimated. Both edges are arithmetic you can redo yourself on numbers already on the record.
Where the market’s half of the subtraction comes from
A market line is a number, but it isn't automatically a probability. The price attached to a side, say −110, converts with its own formula: for a minus price, probability = |price| ÷ (|price| + 100), and for a plus price, probability = 100 ÷ (price + 100). Take the Jaguars' current spread price at DraftKings on this game, −108. That's 108 ÷ 208 = 0.5192, or 51.92%. One book, one side of the spread. It still carries that book's hold, the vig, which is why a single price isn't a fair number yet. Getting to a fair one takes a second step, dividing the hold out against the other side's price, and that's the next piece in this series, worked on a different Week 1 game.
Where our half of the subtraction comes from
We don't run one spread model. We run dozens per game, built different ways, and the one called plainly House NFL Spread is the one that matters here. I checked four separate Week 1 games. Every time, the number that lands in our published pick is that single model's number. It isn't an average of the roster. For Browns at Jaguars, that model says 8.44, and that's exactly what we published. The rest of the models on the same game run from −9.73 all the way up to 8.39. The disagreement underneath is far wider than the one number you get to see.
A full worked example: Browns at Jaguars
| Market | Market line | Our number | Edge (points) |
|---|---|---|---|
| Spread (home, Jaguars) | 7.5 | 8.44 | 0.94 |
| Total | 40.5 | 43.62 | 3.12 |
Read the spread half of that table straight. At the time we made the pick, the market had Jacksonville as 7.5-point favorites. Our own model had them 0.94 points higher, at 8.44. That distance is the edge. 0.94 points between what the market charged and what we thought was fair. The total reads the same way: market at 40.5 combined points, our number at 43.62, a 3.12-point gap on the over.
One thing to flag so you don't cross wires. In our numbers, a positive value on a home-side spread means the home team is favored by that many points. That's the reverse of the −7.5-style notation printed at a sportsbook. Our market line and our own number use the same convention as each other, so the subtraction above stays internally consistent. It just won't line up with a screenshot of a live board until you flip the sign.
What this number does not prove yet
A points edge is a pre-game estimate, not a result. No settled record stands behind this number yet, and the full count is spelled out where the numbers come from, further down this page. A 0.94-point or 3.12-point gap says our model and the market disagree by that much on one game. It says nothing about which of us was right.
It isn't a probability either. Turning a points gap into a probability gap takes an extra assumption: how many percentage points of win probability one point of spread buys near a given number. That conversion, plus the no-vig arithmetic on the market side of it, is the next two pieces in this series. Both are worked on different Week 1 games, with the assumption said out loud instead of folded in quietly.
This pick sits on our Week 1 board with everything else, over on the NFL picks board and the live NFL odds desk. How to read a price before you subtract anything from it is in how to read betting odds and where a book's hold hides.
The Receipts Drawer
Two gaps on one game: 0.94 points on the spread, 3.12 points on the total, both of them plain subtraction with both halves shown to you. Neither one has a win or a loss attached, so what you're holding is a disagreement worth watching, not a result. And the market half of that subtraction moves. If the line shifts before kickoff, the edge shifts with it, so recheck against the live odds desk before you go shopping the number again.
