Sportsbook prices usually imply probabilities that add to more than 100%. The excess is the overround, one way to describe the book's built-in margin. Removing it gives a cleaner market baseline.
You need both sides of the same market from the same snapshot. The odds calculator performs the normalization, but the calculation is short enough to check by hand.
Step 1: convert both prices
For a market priced -110 on each side, the implied probability for either side is:
The raw pair is therefore 52.38% and 52.38%.
Step 2: calculate the overround
Add the two raw probabilities:
The market totals 104.76%, so the overround is 4.76 percentage points.
Step 3: normalize to 100%
Divide each raw probability by the total:
For the -110/-110 market:
The normalized no-vig estimate is 50% for each side.
Uneven-price example
Suppose the favorite is -180 and the underdog is +155.
| Side | Raw implied probability | No-vig probability |
|---|---|---|
| Favorite -180 | 64.29% | 62.11% |
| Underdog +155 | 39.22% | 37.89% |
| Total | 103.50% | 100.00% |
The raw favorite probability is not the clean market estimate. After normalization, the favorite's no-vig probability is about 62.11%.
What no-vig probability can tell you
No-vig probability is useful as a comparison point. It lets you:
- compare a model estimate with a market baseline;
- compare how different sportsbooks split the same market;
- calculate the price needed for a proposed probability;
- avoid counting the book's margin as part of your edge.
What it cannot tell you
Removing the vig does not reveal the true probability. The result still reflects the market's information, timing, liquidity, and pricing choices. A stale market and a current market can produce different no-vig estimates for good reasons.
Simple normalization also assumes the margin is allocated proportionally across both sides. Sportsbooks do not have to distribute their margin that way. More complex methods exist, but proportional normalization is transparent and adequate for a first comparison.
Keep the market snapshot intact
Do not combine the favorite from one time with the underdog from another and call it a two-sided market. Do not mix a moneyline with a spread. Store:
- sportsbook;
- market type;
- both prices;
- observation time;
- game and teams.
Without those fields, the calculation may be correct while the comparison is invalid.
From no-vig probability to edge
Suppose the market's no-vig probability is 62.11% and your model estimates 65%.
That difference is a model claim, not proof of profit. The estimate still needs calibration on future games, and the actual wager must be evaluated at the price you can place. The expected value and Kelly guide covers that next step.