Devigging Explained: How to Strip the Bookmaker Margin and Find True Probability

Every time you place a bet, you pay a hidden tax. It is baked into the odds before you even open your account. The bookmaker calls it the margin; bettors call it the vig or overround. Strip it away and what remains is the market's best estimate of true probability — the number that separates a value bet from a donation. That stripping process is called devigging, and it is the foundation of every serious betting model.

What is the vig (margin / overround)?

Bookmakers price markets so that the implied probabilities of all outcomes sum to more than 100%. The excess is their margin — the cut they keep regardless of who wins. A coin-flip priced fairly would offer 2.00 on each side (50% + 50% = 100%). A bookmaker offering 1.91 / 1.91 collects 52.4% + 52.4% = 104.8% implied — a 4.8% overround. For every €100 staked by the field, the book expects to keep roughly €4.80.

The same structure applies to three-way football markets (1X2), baseball run-line, tennis match winners — any market with a fixed set of exhaustive outcomes. Margins vary wildly: Pinnacle often runs under 2% on major football leagues; a soft book might charge 8–12% on the same match. That gap is where value betting lives.

Worked example: removing the margin from a 2-way market

Suppose Pinnacle prices a tennis match: Player A at 1.72, Player B at 2.20. The implied (raw) probabilities are:

• Player A: 1 / 1.72 = 58.14%
• Player B: 1 / 2.20 = 45.45%
• Total: 58.14% + 45.45% = 103.59% — the margin is ~3.59%.

The multiplicative (proportional) devig divides each raw probability by the total. This scales all outcomes down proportionally until they sum to exactly 100%:

• Player A (de-vigged): 58.14% / 103.59% = 56.13%
• Player B (de-vigged): 45.45% / 103.59% = 43.87%
• Check: 56.13% + 43.87% = 100.00% ✓

The fair (no-vig) price for Player A is therefore 1 / 0.5613 ≈ 1.78. If a soft book is offering 1.85 or higher on the same selection, your expected value is positive — you are getting paid more than the sharpest public probability says you should be.

Worked example: the 3-way 1X2 football market

Football adds a third outcome (draw), which makes the maths look more complex but the principle is identical. Take Pinnacle's closing prices on a Premier League match: Home 2.45, Draw 3.40, Away 3.10. Raw implied probabilities:

• Home: 1 / 2.45 = 40.82%
• Draw: 1 / 3.40 = 29.41%
• Away: 1 / 3.10 = 32.26%
• Total: 40.82% + 29.41% + 32.26% = 102.49% — margin ≈ 2.49%.

Multiplicative devig — divide each by 102.49%:

• Home (de-vigged): 40.82% / 102.49% = 39.83% → fair odds ≈ 2.51
• Draw (de-vigged): 29.41% / 102.49% = 28.69% → fair odds ≈ 3.49
• Away (de-vigged): 32.26% / 102.49% = 31.48% → fair odds ≈ 3.18

Now compare these de-vigged fair prices against what softer books are offering. If Bet365 shows the Away at 3.40 while Pinnacle's de-vigged estimate is 3.18, that is a +6.9% EV opportunity worth investigating. This is the exact logic that underpins value betting.

Why Pinnacle's odds are the best starting point

You can devig any book's prices — but not all starting points are equally informative. A sharp book like Pinnacle accepts large bets from professional bettors, runs margins below 2% on major markets, and adjusts its lines within minutes when informed money arrives. By kick-off its prices have absorbed thousands of sharp bets. The resulting closing line is the most accurate publicly available estimate of true probability available to retail bettors.

De-vigging a soft book's opening lines, by contrast, gives you that book's initial guess before the market corrected it — potentially full of error. This is why sharp vs soft bookmakers is not just a trivia distinction: it determines whether your de-vigged number is a useful signal or noise.

At TheSharpBook we de-vig Pinnacle's closing line as the probability anchor for every market. The de-vigged probability is what we compare against a soft book's available odds to calculate expected value. No Pinnacle anchor, no bet.

From de-vigged probability to expected value

Once you have the de-vigged true probability p, the EV formula is straightforward:

The threshold matters. A 1% EV looks good in theory but disappears under rounding errors, account limits, and small calibration errors in the model. We require a minimum of 5% EV before surfacing a bet, and cap at 18% — anything above that is almost always a data artifact or an extreme outlier where calibration breaks down.

This directly connects to closing line value: if you consistently find positive EV against a sharp de-vigged probability, you will also tend to beat the closing line — which is the most reliable long-run performance signal in betting.

Caveats: the favourite-longshot bias and devig method choice

Two things to keep in mind before treating de-vigged probabilities as ground truth.

First, the favourite-longshot bias: even the sharpest markets consistently underestimate favourites and overestimate longshots. A de-vigged probability of 80% on a heavy favourite may still be 1–2% too low; a de-vigged 8% on a long shot may be 1–2% too high. This does not mean devigging is useless — it means the highest-odds selections (above ~3.50) carry extra uncertainty, which is why we cap our maximum odds at 3.50.

Second, devig method choice matters at the margins. The multiplicative method (used above) is the standard and works well for 2- and 3-way markets. Other approaches — additive devig (subtract margin equally from each implied probability), power method (find the exponent that normalises all probs) — can produce meaningfully different results on lopsided prices. For most retail use cases and balanced markets, the multiplicative method is accurate enough. On very asymmetric markets or long odds, the power method tends to handle the favourite-longshot bias more gracefully.

• Multiplicative: simple, transparent, good default for 2- and 3-way balanced markets.
• Additive: treats all outcomes identically regardless of price — least accurate on lopsided markets.
• Power (Shin): mathematically accounts for longshot bias; best for markets with wide price ranges.

For practical value betting the distinction rarely changes a bet decision by more than a percentage point. Calibrating your own model against historical data, as we do at TheSharpBook, will do far more for accuracy than agonising over devig method selection.

Putting it all together

Devigging is not exotic. It is a two-step calculation: convert each outcome's odds to an implied probability, divide by the total. What you are left with is the market's best estimate — stripped of the house cut. Use a sharp book's de-vigged closing price as your probability anchor. Compare it against whatever a soft book is offering. If the soft book's price implies a higher probability than the sharp anchor says is fair, you have positive EV. That is a value bet.

The statistical model at TheSharpBook goes one layer deeper: we build our own probability estimates from match data across seven sports, then cross-reference them against de-vigged Pinnacle prices to confirm or sharpen the signal. But even without a proprietary model, understanding devigging gives any bettor the foundation to think in probabilities rather than odds — and that alone separates the top 5% from everyone else.