No-vig probability explainedNo-Vig Probability Explained: Separating Market Opinion from Sportsbook Margin

Why the two sides usually add up to more than 100%

In a two-way market priced -110 on both sides, each side carries a raw implied probability of about 52.38%. Added together, those probabilities equal about 104.76%, not 100%. The excess is commonly called the overround, hold, or vig embedded in the posted prices.

That means a single raw implied probability is not a pure statement of what the market believes. It combines the relative weight placed on that outcome with the price the sportsbook is charging. Removing the margin helps separate those concepts.

Normalize all mutually exclusive outcomes

The simplest proportional no-vig method divides each outcome's raw implied probability by the sum of the implied probabilities for every mutually exclusive outcome in the market. In a balanced -110/-110 market, each side normalizes from 52.38% back to 50.00%.

The method is reproducible and useful for two-way moneylines, spreads, and totals. It is also an estimate. More advanced margin-removal methods can allocate the vig differently, especially in heavily asymmetric or multi-way markets, so the chosen method should be stated when precision matters.

Worked example: a -130 favorite and +110 underdog

A -130 favorite has a raw implied probability of approximately 56.52%. A +110 underdog has a raw implied probability of approximately 47.62%. Together they total about 104.14%.

Dividing each side by that total produces no-vig estimates of approximately 54.27% for the favorite and 45.73% for the underdog. Those normalized probabilities add to 100% and provide a cleaner representation of the relative market view.

No-vig probability is a benchmark, not ground truth

Removing margin does not prove that the market is accurate. It creates a normalized baseline reflecting the prices available at a particular time. A model can reasonably disagree because it uses different information, assumptions, injury expectations, starting-lineup projections, environmental inputs, or estimates of team strength.

A large model-to-market gap is not automatically trustworthy either. It can reflect a genuine informational advantage, but it can also reflect stale inputs, an implementation error, an unusual market rule, a data-quality problem, or a model that is poorly calibrated in that situation.

No-vig pricing helps separate two questions: what does the market appear to believe, and what price is the sportsbook charging to express that belief?

Use the normalized market and the offered price for different jobs

The no-vig estimate is useful when comparing the model's probability with the market's relative position. The actual offered price is necessary when calculating the payoff and expected value available to the bettor. Substituting the normalized probability for the actual price would erase the economic cost of placing the wager.

A disciplined workflow can therefore use both: normalize the full market to understand the consensus baseline, then calculate expected value at the exact line and price that can be bet. Recommendation gates can sit above both calculations to decide whether the difference is strong and reliable enough to publish.

• Use both sides of the market when estimating no-vig probability.
• Record the time and source because market probabilities move.
• Calculate bettor EV from the actual offered odds, not the normalized price.
• Treat no-vig output as a reference point rather than a guaranteed fair probability.