Odds are a price from which you can work out the probability the bookmaker builds in. The gap between the "clean" probability and the sum of probabilities across all outcomes is the bookmaker margin.
To judge whether a bet is worth it, knowing who is stronger is not enough. You need to compare your own probability estimate with the one embedded in the odds. Here is how to do it step by step, with numbers.
Converting odds to probability
For decimal odds (the format used on xG.Football) the formula is simple: probability = 1 / odds.
- odds 2.00 → 1 / 2.00 = 50%;
- odds 1.50 → 1 / 1.50 ≈ 66.7%;
- odds 4.00 → 1 / 4.00 = 25%.
The reverse gives fair odds: 1 / probability. If you rate an event at 40%, fair odds are 1 / 0.40 = 2.50.
What the bookmaker margin is
Bookmakers do not price at fair odds: they shorten the odds so that the probabilities of all outcomes add up to more than 100%. That excess is the built-in commission. An example for a 1X2 market:
| Outcome | Odds | 1 / odds |
|---|---|---|
| Home win | 2.10 | 47.6% |
| Draw | 3.40 | 29.4% |
| Away win | 3.60 | 27.8% |
| Total | 104.8% |
A total of 104.8% means a margin of about 4.8%: that is how many extra percentage points are built into the prices.
Removing the margin
The simplest method is to divide each probability by the total (normalisation). In the example above:
- home win: 47.6% / 104.8% ≈ 45.4% (fair odds about 2.20);
- draw: 29.4% / 104.8% ≈ 28.1% (about 3.56);
- away win: 27.8% / 104.8% ≈ 26.5% (about 3.77).
This is a simplification: in practice the margin is spread unevenly and is usually loaded more heavily onto the outsiders' odds. For a quick estimate, normalisation is enough.
Is the bet worth it?
Expected return per unit staked = probability × odds − 1. If you rate a home win at 50% and the odds are 2.10, then 0.50 × 2.10 − 1 = +0.05, or +5% per bet on average over the long run. With a 45% estimate it becomes 0.45 × 2.10 − 1 = −5.5% — a poor bet, even though the event is quite likely.
The key takeaway: value is not about whether the event will happen, but about how far your probability estimate is from the price.
Where to get your own probability
- from your own model or a calculation based on statistics — how the xG.Football model works is described in how the prediction model works;
- from chance-quality metrics — see what is xG;
- from ready-made outcome probabilities on the football predictions page. How closely such probabilities matched reality is visible on the prediction statistics page.
Applying it to other markets
The same logic works for any set of outcomes: the two outcomes of the both teams to score market, totals lines, double chance. Always add up the probabilities of one market's outcomes to see the margin, and compare prices across bookmakers: the odds on a match page are shown per bookmaker.
Common mistakes
- Taking one outcome's probability without accounting for the margin and comparing it with a clean estimate.
- Ignoring small price differences: 1.95 versus 2.00 is a gap of almost 1.3 percentage points.
- Confusing the probability of an event with the value of a bet: a favourite at 1.20 can be underpriced or overpriced.
- Trusting a single estimate: every probability is an approximation; the bigger the gap between your estimate and the price, the sturdier the conclusion.
Unfamiliar terms are explained in the football betting glossary.
Quick recap
| Question | Answer |
|---|---|
| How to convert odds to probability | 1 / odds |
| What is the margin | The sum of all outcome probabilities minus 100% |
| How to remove the margin | Divide each probability by the market total |
| How to tell if a bet is worth it | By your estimate, probability × odds − 1 should be above zero |