Football predictions and bookmaker margins How much is your bookmaker really taking?

A decimal odd represents what €1 staked returns if the bet wins, stake included. An odd of 1.68 returns €1.68 for €1 staked. The lower the odd, the more likely the bookmaker considers the outcome. The inverse of the odd (1 / odd) estimates the implied probability — but it always includes the bookmaker margin, so it is slightly above the true probability.

The implied probabilities of a match's three outcomes always add up to more than 100%: the excess is the bookmaker margin, its commission on every bet. Full calculation, examples and thresholds: our guide to the bookmaker margin.

For each match, MARGE compares the odds of all available bookmakers and finds the cheapest outcome: the market favourite. If even the best odd stays high (above about 2.30), there is no clear favourite — the match is flagged as "balanced" and not turned into a verifiable prediction. Otherwise the read shown is "Favourite win", then refined with team form and standings when that data is available.

The confidence percentage shown on each analysis is not an opinion: it is the implied probability of the favourite, recomputed after removing the bookmaker margin (the three outcomes brought back to 100%). A match at 65% therefore means the market, margin removed, estimates the favourite's win at about two chances in three.

The win rate is the share of winning predictions among those that were settled — cancelled or refunded matches are neutral. ROI assumes the same one-unit stake on every prediction: a +7% ROI means that staking €1 each time would have ended with 7 cents net profit per bet on average. It is more honest than a plain win rate, since it accounts for the odds played.

“Our read” only highlights bets estimated at 70% or better. A high rate is therefore normal and proves nothing on its own. The real question is this: does the probability announced before the match match what actually happened? An achieved rate above the called one points to a cautious model, below it to an overconfident one. On a small sample, a gap of a few points still means nothing.

The headline figure above covers “Our read”, which includes goal markets. Here is the same calculation restricted to outright wins (1X2), whose odds always come from a bookmaker. The two are not directly comparable: a higher hit rate at shorter odds does not mean more profitable. The ROI is what settles it.

A simulation of our “Odds” method on matches already played, using the real results and closing odds (source football-data.co.uk). This is NOT the live timestamped ledger: it is a historical test of the method. Past performance never guarantees future results.

Differences between leagues (a few ROI points) sit within the statistical margin of error: they do not support calling any league “profitable”. Only the global result, across all bets, is significant.