EV Calculator

Say your model makes the Chiefs 55% to cover −3, and the book offers −110. Run the numbers on a $100 stake (an illustrative amount — the percentages are what matter):

• At −110, a winning $100 bet returns $90.91 profit (100 × 100/110).
• EV = (0.55 × $90.91) − (0.45 × $100) = $50.00 − $45.00 = +$5.00 per $100, or +5.0%.

Now the same 55% estimate at −120, the price you often see after a line moves: profit on a win drops to $83.33, so EV = (0.55 × $83.33) − (0.45 × $100) = $45.83 − $45.00 = +$0.83 per $100. One dime of extra juice erased 83% of the edge. This is why line shopping is not a nicety — at typical edge sizes, the price is most of the bet.

All of the arithmetic above is exact formula math, not a historical claim: EV = p × profit − (1 − p) × stake.

American odds convert to a break-even win rate with one formula: for negative odds, break-even = |odds| / (|odds| + 100). At −110 that is 110/210 = 52.38%. Every extra nickel of juice raises the bar by roughly a point:

Exact formula math: break-even = |odds|/(|odds|+100); the +2% column solves p × (100/|odds|×100) − (1−p) × 100 = 2.

The calculator will happily accept “60%” as your win probability on a spread bet. The bundled game log argues for humility. Across 7,276 completed games from 1999–2025, favorites covered the closing spread just 48.7% of the time and the closing line missed the actual margin by 10.3 points on average — single games are extremely noisy, and the market's number is already close to unbiased. Moneyline prices tell the same story: in every price band since moneylines appear in the data (2006), favorites won less often than their price implied.

Data: nflverse games.csv bundled with this site (completed games with moneylines, 2006–2025), computed by the author.

The lesson for this calculator: if your input says you beat closing prices by 5% or more per bet, the most likely explanation is that the input is wrong. Sustained edges of 1–3% are what strong bettors report, which is why closing line value — not short-term results — is the sanity check that matters.

The hardest input in the calculator is the only one that matters. Before typing a win probability, anchor it against what closing spreads have historically implied about the moneyline — computed from the bundled nflverse game log (7,276 completed games, 1999–2025):

Data: nflverse games.csv bundled with this site, computed by the author.

If your moneyline probability for a 3-point favorite is far from 60%, or for a touchdown favorite far from 76%, the burden of proof is on your model, not the market. The same logic applies to spread bets: the calculator's answer is only as honest as the daylight between your number and these base rates.

• EV output is only as good as your probability input. The calculator does exact arithmetic on an estimate; it cannot tell you whether 55% is realistic for the bet in front of you.
• Historical rates above are pooled averages. The moneyline table describes bands of thousands of games, not any single matchup, and band composition changes over time.
• Positive EV does not mean low risk. A +5% edge still loses 45% of the time in the example above; bankroll sizing (see the Kelly calculator) is a separate decision.
• Prices move. The odds you compute EV on must be the odds you actually get.
• The calculator prices one bet in isolation. A slate of positive-EV bets on correlated games carries more variance than the per-bet numbers suggest; size simultaneous positions with the Kelly calculator and its Sunday-card caveats in mind.