How to Remove the Vig From Betting Odds to Find Fair Prices

Implied probability puts every betting line on the same percentage scale, making the sportsbook’s margin easier to spot. A refresher on how sportsbook odds and prices work can help before applying the formulas.

For American odds:

• Positive odds: 100 / (odds + 100)
• Negative odds: |odds| / (|odds| + 100)

At +150, the calculation is 100 / (150 + 100) = 0.40, or 40% implied probability.

At -200, it is 200 / (200 + 100) = 0.6667, or 66.67%.

Convert every possible result in the market, then add the percentages. A two-way moneyline requires both teams; a three-way soccer market requires the home win, draw and away win. The total above 100% represents the market’s built-in vig before normalization.

For a standard two-way market priced at -110/-110, each side carries an implied probability of:

110 ÷ (110 + 100) = 52.38%

Adding every possible outcome exposes the overround:

• Side A: 52.38%
• Side B: 52.38%
• Total implied probability: 104.76%
• Amount above 100%: 4.76 percentage points

A fair market totals 100%. Here, the extra 4.76 points represent the embedded sportsbook margin in the quoted odds. That excess must be removed before estimating each side’s no-vig probability.

Overround should not be confused with actual sportsbook revenue. The difference between sportsbook hold and vig matters because realized hold depends on how much money is wagered on each side, which bets win, line movement, promotions and other operating factors.

The simplest no-vig method is proportional normalization. Divide each outcome’s implied probability by the sum of all implied probabilities:

Fair probability = outcome implied probability ÷ market total

For a standard -110/-110 point spread, each side carries an implied probability of 52.38%. The total is 104.76%, as found when bettors calculate vig in a two-way market.

• Side A: 52.38% ÷ 104.76% = 50.00%
• Side B: 52.38% ÷ 104.76% = 50.00%
• Adjusted total: 100%

Both sides therefore have fair odds of +100. In this balanced example, normalization removes the 4.76-percentage-point overround evenly.

The assumption behind the method

Proportional normalization assumes the vig is distributed according to each outcome’s listed implied probability. That is a useful estimate, but it may not reflect how the sportsbook actually shaded the betting line for expected action or risk.

In an uneven market, dividing by the total is preferable to subtracting the same number of percentage points from every outcome. It preserves the market’s relative pricing while producing a clean 100% total.

Once the market has been normalized, each no-vig probability can be converted into a familiar American betting line. Let p represent the fair probability as a decimal.

• Below 50%: Fair odds = 100 × (1 − p) ÷ p
• Above 50%: Fair odds = −100 × p ÷ (1 − p)
• Exactly 50%: fair odds are +100 (even money)

For example, a 42.86% probability becomes:

100 × (1 − 0.4286) ÷ 0.4286 = +133.32

Its 57.14% counterpart becomes:

−100 × 0.5714 ÷ (1 − 0.5714) = −133.32

Rounding both to +133 and −133 is sensible. Whole-number rounding keeps the result readable without implying more precision than the original betting market supports.

These fair odds are a comparison benchmark, not necessarily a wager available at a sportsbook. If the no-vig price is +133 and a book offers +140, the available line is more favorable on price alone; other market and bankroll considerations still apply.

Consider a two-outcome moneyline with one side at -150 and the other at +130.

1. Convert both odds

• -150: 150 ÷ (150 + 100) = 0.6000 → 60.00%
• +130: 100 ÷ (130 + 100) = 0.4348 → 43.48%

Together, the implied probabilities equal:

60.00% + 43.48% = 103.48%

The extra 3.48 percentage points represent the market’s overround.

2. Normalize the probabilities

Divide each implied probability by the full 103.48% total:

• Favorite: 60.00 ÷ 103.48 = 57.98%
• Underdog: 43.48 ÷ 103.48 = 42.02%

These no-vig probabilities now total 100%.

3. Convert back to fair odds

• 57.98%: -(57.98 ÷ 42.02) × 100 ≈ -138
• 42.02%: (57.98 ÷ 42.02) × 100 ≈ +138

The estimated fair prices are therefore -138 and +138.

Simply subtracting 3.48 points from each side is incorrect. That produces 56.52% and 40.00%, totaling only 96.52% because the entire overround was removed twice. Splitting the overround equally would total 100%, but it assumes the juice was added evenly. Proportional normalization instead preserves each outcome’s share of the original market, making it the cleaner default for a basic no-vig calculation.

Three-way markets use the same proportional normalization method, but every possible outcome must be included. Consider a soccer 1X2 moneyline offering home, draw, and away prices:

Outcome	Posted odds	Implied probability	Normalized probability	Fair price
Home	+120	45.45%	42.84%	+133
Draw	+240	29.41%	27.72%	+261
Away	+220	31.25%	29.45%	+240
Total	—	106.12%	100.00%	—

Each implied probability is divided by 106.12%, the market’s full implied total. The resulting probabilities are then converted back to American odds. Small differences may appear when percentages or fair prices are rounded.

Leaving out the draw would produce a misleading result because the home and away prices were set as parts of a three-outcome market. The same caution applies to any moneyline offering a tie as a separate result.

Proportional devigging removes the visible overround; it does not prove that the remaining probabilities reflect each outcome’s true chance. The method assumes juice is allocated proportionally, but a sportsbook may shade a popular NFL favorite, protect against one-sided action, or place more margin on a longshot.

The estimate becomes less dependable when:

• The betting line is stale and has not reacted to injuries, weather, or market movement.
• The market is illiquid, as can happen with obscure props, lower-tier leagues, or early openers.
• Vig is distributed unevenly between outcomes rather than applied at a consistent rate.

Other methods can produce different fair prices. An additive model removes equal percentage points from each outcome, while a power model adjusts favorites and longshots differently. Those differences are usually more noticeable in three-way markets and prices with heavy favorites.

Rather than treating one sportsbook as the source of truth, it is stronger to build a consensus price across several sportsbooks. Compare current lines with identical rules and settlement terms, giving more weight to active, liquid markets. Even then, the result is a benchmark—not proof that a wager has value.