Calculator

No-Vig Calculator

About this calculator

By Alex Thornton · author of the MIT-licensed teachersbettextbook library · Last reviewed 24 July 2026

The price you see has the bookmaker's cut baked into it. Take the cut out and you get the real probability underneath.

Full animated derivation, step by step: open the De-Vig tool in the app →

A no-vig calculator strips the sportsbook’s margin out of a two-sided price and returns the fair probability of each side. It is the first step of every edge calculation on this site.

What the vig actually is

Add up both sides of any real market and you get more than 100%. That is not a rounding error. A book quoting a true coin flip at even money either way would earn nothing, so it shades both sides instead.

A normal two-way NFL moneyline carries about 4.5 extra points of probability. That is the cut, and you pay it whichever side you take.

Strip it out and you can finally compare things. A raw 60.3% and a fair 58.9% are not the same number, and everything else here (expected value, Kelly, closing line value) starts from the fair one.

Three methods do the stripping and they disagree with each other. Pick one in the form. The argument is below.

Want more math?the Nerd Corner

American odds convert to a vig-inclusive implied probability like this:

q = \dfrac{100}{100 + m}\ \ (m > 0), \qquad q = \dfrac{-m}{100 - m}\ \ (m < 0)

The two sides sum to more than one. That excess is the vig, and the simplest way to remove it is to divide each side by the total:

v = q_A + q_B - 1, \qquad p_i = \dfrac{q_i}{q_A + q_B}

where:

$m$ : the posted American moneyline for one side (e.g. −152 or +138), unitless.
$q_A,\ q_B$ : the vig-inclusive implied probabilities of side A and side B, each between 0 and 1. They sum to more than 1 on any real book.
$v$ : the vig, also called the overround or hold, in probability points. A typical two-way NFL moneyline runs 2–5%.
$p_i$ : the fair, no-vig probability of side $i$ . By construction $p_A + p_B = 1$ .

Worked example: the numbers in the form below

-152 \Rightarrow q_A = \tfrac{152}{252} = 60.32\%, \qquad +138 \Rightarrow q_B = \tfrac{100}{238} = 42.02\%

v = 60.3175\% + 42.0168\% - 100\% = 2.3343\% \approx 2.33\%

p_A = \dfrac{0.60317}{1.02334} = 58.94\%, \qquad p_B = \dfrac{0.42017}{1.02334} = 41.06\%

The fair line is −143.6 / +143.6. The price called side A a 60.3% favorite; with the book’s cut removed it is a 58.9% favorite, 1.4 points lower.

Proportional vs power vs Shin: which method to use

All three methods rescale the same two implied probabilities to sum to one. They differ in where the margin comes from.

Proportional divides each side by the total, so a side carrying twice the implied probability gives back twice the margin:

p_i = \dfrac{q_i}{q_A + q_B}

Power raises each implied probability to a common exponent $k$ , solved so the results sum to one. Because $q^k$ shrinks a small number proportionally harder than a large one, power takes more margin out of the longshot:

q_A^{\,k} + q_B^{\,k} = 1, \qquad p_i = q_i^{\,k}, \qquad k \ge 1

Shin models the book as padding its prices against a fraction $z$ of insider money, and solves for the $z$ that makes the fair probabilities sum to one:

p_i = \dfrac{\sqrt{z^2 + 4(1-z)\,q_i^2 / s} - z}{2(1-z)}, \qquad s = q_A + q_B

One market carried end to end through all three, −250 / +205, a 4.22% hold:

q_A = \tfrac{250}{350} = 71.43\%, \qquad q_B = \tfrac{100}{305} = 32.79\%, \qquad s = 104.22\%

Method	Fitted parameter	Fair favorite	Fair longshot	Fair line
Proportional	none	68.54%	31.46%	−218 / +218
Power	k = 1.0716	69.73%	30.27%	−230 / +230
Shin	z = 4.25%	69.32%	30.68%	−226 / +226

The longshot moves 1.19 points across the three methods, a bigger gap than most of the edges people hunt for. The spread scales with the favorite: at −110 / −110 all three return exactly 50.00%, since a symmetric line has nowhere asymmetric to put the margin, while at −2000 / +900 the longshot’s fair price runs from +952 (proportional) through +1255 (Shin) to +1612 (power). There the method is most of the answer.

Which method is correct is not settled here, because the fair probability is never observed. Power and Shin shade the longshot down, the direction settled-market studies of favorite–longshot bias point; proportional is the neutral baseline. Reading all three and treating the spread as your uncertainty is more honest than reporting one to four decimals. A tool that prints a single number without naming its method is hiding its largest source of error.

Where the de-vig fits, and what goes wrong

It turns a reference line into a probability. A posted price is an estimate inflated by that book’s margin; de-vigging recovers the estimate underneath. But de-vigging a soft book tells you what that book thinks, not what the market thinks: the result is only as good as the line fed in.
It feeds the rest of the math. Expected value compares a de-vigged fair probability to what a different price implies; closing line value is the closing fair probability minus the probability your price implied; the vig calculator reports the hold on its own.
Both numbers have to be on the same footing. Comparing a 58.94% fair probability to a raw 60.32% implied probability manufactures a 1.4-point gap out of nothing but the vig. Two books quoting different holds are not comparable until both are de-vigged.
Normalizing is not averaging. Halving the overround and subtracting it from each side is a different operation, and it disagrees with proportional normalization on any market that is not symmetric.
The fair price is an estimate, not a quote. Nobody is offering −143.6.
Proportional is a choice, not a neutral default. It leaves the longshot with the highest fair probability of the three methods, which on a −2000 line is worth several hundred points of price.

Check this number yourself

The page runs the same math as the MIT-licensed teachersbettextbook package: devig(ml_a, ml_b, method), with the vig from overround(ml_a, ml_b).

pip install teachersbettextbook

>>> from teachersbettextbook import devig, overround
>>> overround(-152, 138)
0.023342670401493848
>>> devig(-152, 138, method='proportional')
(0.5894160583941606, 0.41058394160583944)

Same inputs, same digits, no account required. More on what is open and what is not: open source.

The de-vig is step one for a +EV bet, a sharp-consensus fair line, and closing line value. Full method and animated derivation: how to de-vig odds.

Frequently asked questions

What is a no-vig calculator?

A no-vig calculator removes the sportsbook’s built-in margin (the vig or juice) from a two-sided price, leaving the fair probability and fair odds the line implies. You enter both sides; it returns the de-vigged probabilities that sum to 100%.

How do you remove the vig from odds?

Convert both sides to implied probabilities, then normalize so they sum to one. The simplest way is proportional (divide each by the total); the power and Shin methods shade the favorite and longshot differently to correct for favorite-longshot bias.

What is the vig?

The vig (vigorish, or juice) is the sportsbook’s margin, baked into the price so the two sides’ implied probabilities sum to more than 100%. That overround is the house edge on the market.

Which de-vig method is the most accurate?

No method is right by definition, because the fair probability is never observed. Proportional spreads the margin in proportion to each side; power and Shin take more out of the longshot, which is what settled-market studies of favorite-longshot bias tend to show. The three agree exactly at −110 / −110 and diverge by 1.48 points at −300 / +240, so the honest move is to read the spread between them as your uncertainty.

Can you remove the vig from a three-way market?

The same normalization works for any number of outcomes: convert every price to an implied probability and rescale the set so it sums to one. This calculator is deliberately two-way, because the rest of the math on this site is defined on two-outcome markets.