Devigging odds: removing the bookmaker margin
Four methods, when to use each, and the trap of using only one.
TL;DR
Bookmakers price markets so their implied probabilities sum to more than 100%. Devigging is the process of stripping that margin so the prices sum to exactly 100% — giving you a "fair" probability estimate. There are four common methods, each with different bias profiles.
Why we devig
Implied probability from a decimal odd is just 1 / odds. A 3-way market like 1X2 typically sums to 102–108%. The 2–8% overround is the bookmaker margin (a.k.a. vig, juice, overround).
To find +EV (see +EV guide) you need to know what the market *truly* believes — not what it advertises. Devigging gives you a fair-probability estimate.
The four methods
1. Multiplicative (proportional)
The simplest. Divide each implied probability by the sum.
imp_home = 1 / 1.90 = 52.6%
imp_draw = 1 / 3.60 = 27.8%
imp_away = 1 / 4.50 = 22.2%
sum = 102.6%
fair_home = 52.6% / 1.026 = 51.3%
fair_draw = 27.8% / 1.026 = 27.1%
fair_away = 22.2% / 1.026 = 21.6%Trade-off: assumes the margin is distributed proportionally. Reality: bookmakers systematically build *more* margin into favourites and longshots. So this method underestimates favourites and overestimates longshots. Cheap to compute, biased at the extremes.
2. Additive
Subtract the per-leg margin equally from each implied probability.
margin_per_leg = (102.6% - 100%) / 3 = 0.87%
fair_home = 52.6% - 0.87% = 51.7%
fair_draw = 27.8% - 0.87% = 26.9%
fair_away = 22.2% - 0.87% = 21.3%Trade-off: works fine for balanced markets but breaks at extreme prices (subtracting a fixed amount from a small probability disproportionately impacts it).
3. Power
Adjust the implied probabilities by raising to a power and rescaling so they sum to 100%. The power is found by iteration.
Trade-off: Handles the favourite-longshot bias well. More compute, no clean closed form. Default choice for 2-way markets (BTTS, totals).
4. Shin
Models the share of "insider" bettors in the market. Solves a small quadratic equation to estimate the insider proportion z, then derives the fair probabilities.
Trade-off: statistically the most accurate for high-margin markets (1X2 and player props). Slight extra compute. Default choice for 1X2.
Which to use
| Market | Recommended method |
|---|---|
| 1X2 (3-way h2h) | Shin |
| 2-way h2h (Draw No Bet) | Power |
| Totals (Over/Under) | Power |
| Both Teams to Score | Power |
| Player props | Power or Shin depending on margin |
The trap: only ever using multiplicative
Multiplicative devig systematically misprices favourites and longshots. If your +EV scanner uses only multiplicative on a market with significant favourite-longshot bias (1X2, props), you'll overestimate EV on longshots and underestimate it on favourites. Real money has been lost on this exact bias.