NFL Win Probability Calculator
Spread, total, or a contract price → win probability, the market\u2019s implied number, and the gap between them
win_prob(home) = Φ( spread_home / 13.29 )
where spread_home is the home team's expected margin (a −6.5 favorite = +6.5), Φ is the standard normal CDF, and 13.29 is the spread of results around a market line. Start from two teams instead and it widens to 14.4, because our own projected margin is the less accurate centre.
Home favored is negative — e.g. −6.5 means home by 6.5.
Win probability from the scoring-margin model (σ = 13.29, the spread of results around a market line). Projected score splits the total by the spread. An efficient market line is ~50/50 against the spread — the edge is in disagreeing with it.
Use the NFL Win Probability inside your AI agent
This tool is also live on our MCP server as nfl_win_probability. Connect Claude, ChatGPT, Cursor, or Copilot and call it directly — six core calculators are free.
Preseason offer
Gridiron Edge Pro — 50% off your first payment
Pro opens every engine on the site — the cross-market arb scanner, the Mispricing Scanner, Thee Oracle, and the silver, gold, oil and bitcoin edge models that price commodity contracts in real time.
$75 for the first year, or $7.50 for the first month. Code applies automatically. Ends Sep 10.
Every signal graded in public, losers left on the board. See the record →
Related Tools
A point spread is really a prediction of the final margin. This calculator turns that margin into a win probability the same way the sharpest models do — through a normal distribution around the expected margin. The wider the spread, the more confident the favorite; but NFL games are high-variance (a13.29-point standard deviation), so even a touchdown favorite loses outright almost a third of the time.
The projected score splits the game total by the spread, and the cover/over probabilities are measured against the line — near 50% when the line is efficient. The real edge comes from an independent power rating that disagrees with the market, which is what the Gridiron Edge model publishes across the /nfl pages and the nfl_edge MCP tool.
Frequently Asked Questions
How do you convert an NFL spread into a win probability?
A market spread already is the market's own projected margin, so it runs through a normal distribution with the spread of results around that centre — σ = 13.29 points, measured on 5,695 games from 2003 to 2024. Win probability = Φ(margin / 13.29). A 6.5-point favorite wins about 69% of the time; a 3-point favorite about 59%. Enter two teams instead of a spread and the calculator uses OUR projected margin, which is less accurate than a market line, so it widens to σ = 14.4 and returns slightly lower confidence.
What do the projected scores mean?
The game total is split by the spread: home = total/2 − spread/2, away = total/2 + spread/2. So a total of 47.5 with the home team favored by 6.5 projects roughly 27–20.
Why is the cover / over probability around 50%?
An efficient market spread and total ARE the expected margin and the expected points, so against the line itself both are close to a coin flip. The edge only appears when an independent model disagrees with the line — which is what the Gridiron Edge model does on the /nfl pages.
Is this the same model behind the PredictionMarketsPicks NFL edges?
Yes — the win-probability math and the σ constant are ported directly from the Gridiron Edge engine that powers our NFL power ratings, win-total futures, and Kalshi edge board. It is also callable by AI agents as the nfl_win_probability MCP tool.
Get The 7 Oracles' daily edge — subscribe free
No spam. Unsubscribe anytime.