De-vigging, defined
De-vigging removes the vig from a set of odds, so the implied probabilities sum back to exactly 100%.
, 3 min read
Reviewed by Phil McParlane, Founder, on
The simplest method — proportional normalisation — divides each outcome's implied probability by the market total. If a market implies 55% and 50% (105% together), the de-vigged fair probabilities are 55 / 105 = 52.4% and 50 / 105 = 47.6%.
What remains is the market's genuine estimate, stripped of margin, and therefore directly comparable against your own model or against another market. The gap between the raw total and 100% is the overround — the margin itself — and de-vigging is the step that removes it, so two markets priced with different margins can be read on one scale. That "sums to 100%" property is not just theory: across the 59 CS2 match markets our line served on 16 July 2026, the de-vigged implied probabilities summed to exactly 100.0000%, the margin fully removed, every time.
Proportional normalisation is the workhorse. It assumes margin falls on each outcome in line with its probability. Other methods split it another way. With a heavy favourite, they give the longshot a lower fair probability. The table below shows the gap. Some models go further. To compare ways of turning odds into forecasts, read Štrumbelj (2014).
No-vig odds, fair odds and fair value
No-vig odds and fair odds both mean de-vigged prices, with no margin on top. A price like that breaks even only if the probability behind it is right.
For example, a market at −110 on both sides implies 52.38% each (110 ÷ 210, see implied probability) and 104.76% together. De-vigged, each side is 50%, or +100.
"Fair value" has two main uses. From a market, it is those de-vigged prices, so it carries whatever the market believes. From a model, it is your own probability turned into a price: fair decimal odds = 1 ÷ your probability. So 50% is 2.00, or +100, and 25% is 4.00, or +300.
The method changes the answer
Proportional normalisation is also called the multiplicative method. Three others are additive, Shin and power. All four agree on an exactly even (50/50) market, but not when one side is a heavy favourite.
For example, −600 and +425 imply 85.71% and 19.05%, a total of 104.76%. The table shows the longshot's fair probability by method.
| Method | Longshot's fair probability | Against multiplicative (points) |
|---|---|---|
| Multiplicative | 18.18% | 0 |
| Additive | 16.67% | −1.52 |
| Shin | 16.67% | −1.52 |
| Power | 15.77% | −2.41 |
The widest gap is 2.41 percentage points. The last column uses unrounded figures. Additive and Shin match on any two-way market. See four de-vig methods compared for how each works.
What no-vig odds and fair value are not
They are not a prediction or an edge. They are one reading of the market's view with the margin taken off. Whether that view is right is a separate question. A model's fair value is only as good as the probability it came from. When it differs from the de-vigged market price, the two disagree and which one is wrong is an open question. A gap is not a recommendation.
From one market to a whole schedule
A derived market line covers many markets at once. For each one, it takes the vig out of each source, finds the median, then rescales it to 100%. We never name a source or serve a per-book price. Only the combined line is published.
Read the methodology behind the served line, or the vig explained: how de-vigging finds fair odds for the arithmetic and its limits.
To do it on your own numbers, the free no-vig calculator removes the margin by four different methods — multiplicative, additive, power and Shin — side by side.
Start free with a sandbox key: a fixed sample of past matches, no card.