No Vig Fair Odds Calculator: Three Methods, Worked Through

How to remove vig from odds: the three-step conversion

Removing vig is three steps: convert each price to an implied probability, add them up to measure the margin, then shrink the probabilities back to 100% using one of the methods below. Only the third step differs between calculators.

Take the two-way line from the Python shin package's documented example: decimal 1.5 on the favorite and 2.74 on the underdog. In American terms that is -200 and +174.

That 3.16% overround is what every method has to remove. The question each one answers differently is who pays it: the favorite, the underdog, or both in proportion.

Multiplicative method: divide by the total

The multiplicative method divides each implied probability by the sum of implied probabilities, so every outcome loses the same share of its own size. The CRAN implied package describes it as dividing the inverted odds by their sum and calls it the basic method.

Worked through on the example line: 66.67% / 103.16% = 64.62% for the favorite and 36.50% / 103.16% = 35.38% for the underdog. Inverted, those are fair decimal odds of 1.547 and 2.827, or roughly -183 and +183 in American.

It is the simplest version, one division per outcome. Its known weakness is bias: normalization ignores the favorite-longshot bias, the tendency for long shots to carry more of the margin than favorites. The same vignette ranks it as the least accurate of the methods it ships.

Additive method: subtract an equal slice

The additive method takes the margin, splits it evenly across outcomes and subtracts the same amount from each. The vignette spells out an equal amount of the margin per outcome.

On the example line the 3.16% margin splits into two slices of 1.58%. The favorite drops from 66.67% to 65.09% and the underdog from 36.50% to 34.91%. Fair odds come out at 1.536 and 2.864, roughly -186 and +186.

Those two probabilities match the shin package's output for the same odds, 65.09% and 34.91% under Shin, to every decimal place shown. That is not a coincidence. Both the Clarke, Kovalchik and Ingram paper and the implied vignette state that Shin and additive are equivalent for two competitors. Shin's model assumes a share of bettors are insiders and the book sets prices to protect itself, and with two outcomes that insider share has a closed-form analytic solution (about 0.0317 here). The practical read: a calculator offering both "Shin" and "additive" on a two-way market is offering the same number twice.

Additive has a failure mode on longer markets. The vignette warns that the additive method can produce negative probabilities, and the paper lists the same defect. A long shot whose implied probability is smaller than its equal slice of the margin goes below zero, which no fair price can be.

Power method: raise to a fixed exponent

The power method raises every implied probability to the same exponent k, choosing k so the results sum to exactly 100%. The vignette frames it as modelling bookmaker probabilities as a power function of the proper probabilities.

On the example line, the exponent works out to about 1.0505. That gives 65.32% for the favorite and 34.68% for the underdog, fair odds of 1.531 and 2.883, roughly -188 and +188. Because raising a small number to a power above one shrinks it proportionally more, the underdog gives up more of the margin than under either other method. That is the favorite-longshot adjustment built into the formula.

The paper notes that powers never leave the 0-1 range, so power cannot produce the negative probabilities additive can. On accuracy, the authors found across three bookmaker datasets that power universally outperforms the multiplicative method and is comparable to or better than Shin.

Where the methods disagree, side by side

All three methods agree on the margin and disagree only on who absorbed it. On a tight two-way line the spread is small; it widens as the margin grows and as the price gap between favorite and underdog widens.

Read the table as a range, not a verdict. If you are holding the underdog at a price you think beats fair, multiplicative is the most generous benchmark and power the strictest. A bet that only clears the multiplicative line is a thinner edge than one that clears all three.

On a three-way market (home, draw, away) the same three steps apply, but Shin stops matching additive, additive starts risking that negative-probability failure on a big outsider, and the gaps between methods grow. Those are the conditions where the method choice changes the fair price most.

How to check any no-vig odds calculator

Feed a calculator a two-way line and compare its output to the three rows above. If the fair favorite probability equals the implied probability divided by the total, it is multiplicative. If both sides moved down by the same number of percentage points, it is additive, and any "Shin" toggle on that screen will return the identical result. If the underdog lost proportionally more than the favorite, it is power.

Recent work treats these as the standard toolkit: a 2026 arXiv study names Multiplicative, Shin and Power as the existing odds-only methods and tests them on 90014 football matches across five bookmakers. None of them turns a fair price into a forecast you can count on. They tell you what the market's number is with the bookmaker's cut taken out, and that is information, not a promise.

Market questions

What does a no vig odds calculator actually remove?

It removes the bookmaker margin, the amount by which the implied probabilities of all outcomes add up to more than 100%. What it puts back is a fair price under one assumption about how that margin was spread, which is why the method setting matters.

Is Shin's method better than additive on a two-way line?

No difference exists to find. With only two outcomes the two methods are equivalent, so they return the same fair probabilities. Shin only separates from additive once a market has three or more outcomes.

Why can additive de-vigging give a negative probability?

Additive subtracts the same slice of margin from every outcome. A long shot whose implied probability is smaller than that slice ends up below zero, a failure the power method avoids because it never leaves the 0-1 range.

Which method should I trust on a three-way or multi-runner market?

The research leans toward power. In the Clarke, Kovalchik and Ingram comparison it beat the multiplicative method on every dataset and matched or beat Shin, while additive can produce negative probabilities, especially with many outcomes.

Price and data sources

• American Journal of Sports Science (Clarke, Kovalchik, Ingram): Adjusting Bookmaker's Odds to Allow for Overround, 2017-12-25
• CRAN (Jonas C. Lindstrom): Introduction to the implied package, 2023-06-11
• arXiv (Goto, Takeishi, Yairi): Forecast Sports Outcomes under Efficient Market Hypothesis: Theoretical and Experimental Analysis of Odds-Only and Generalised Linear Models, 2026-04-19
• GitHub (mberk): shin: Python implementation of Shin's method, accessed 2026-10-16