What value means
A value bet is one where your estimate of the chance is higher than the one built into the odds. If you are sure a team has a 50% chance and the bookmaker offers 2.30 (about 43%), the bet returns 15% on average — if you are right. The whole difficulty lies in that “if”.
It is the only strategy with any chance of profit at all: all the others just reshuffle the margin. But it needs an estimate of the chance that is more accurate than the market's.
How we tested our model
The Forepitch model was tested the same way as on the accuracy page: it learnt only from past matches and estimated the next ones as if they had not been played yet. From 27 January 2023 to 12 October 2026 that makes 22,108 matches with odds. In each match we took the outcome where the model's chance times the early odds (a day or two before the match) is greater than one plus the edge, and staked 1 unit.
| Model edge | Bets | Average bookmaker | Best price | Beat the close |
|---|---|---|---|---|
| 0%+ | 9,553 | −10.3% | −3.3% | 44% |
| 5%+ | 6,026 | −13.7% | −6.9% | 45% |
| 10%+ | 3,811 | −18.8% | −10.5% | 44% |
- Average bookmaker
- Best price on the market
The result is negative, and the bigger the edge the model sees, the worse it gets: from −10.3% to −18.8% at the average price. So where the model disagrees with the market, the market is more often right. By the average error of the probabilities (log loss: the lower, the better), the sharp line's closing price is more accurate than the model: 0.9932 against 1.0088.
The model is well calibrated, though: outcomes it rates at 60% happen about 60% of the time. But betting needs more than an honest probability — it needs a probability more accurate than the market's.
Why the market has the upper hand
The closing price pools the knowledge of thousands of people, including professionals with big stakes: line-ups, injuries, motivation, weather. The model sees the statistics and does not know what came out an hour before kick-off. To beat the market consistently, you need to know something it has not taken into account — and act before the price changes.
CLV: how to tell whether you have an edge
The best sign of an edge is CLV (closing line value): whether you got higher odds than the price at kick-off. Betting results are very noisy: even over thousands of bets, luck moves them by a few per cent. CLV shows up much faster, because it compares a price with a price, not with an outcome.
Our model's value bets at the best early price have a CLV of −1.4%, and 44% of them beat the close: on average they are worse than the closing price. For comparison, the article on early odds has a strategy with a positive CLV — there the edge comes not from a model but from the sharp line itself.
The bottom line
- Value betting is the only strategy with a chance of profit, but only if your estimate of the chance is more accurate than the market's.
- Check your edge by CLV over hundreds of bets, not by one lucky week.
- Our model's probabilities are honest, but the closing market is more accurate, and the model has no edge over it yet. In Premium the model runs value bets live — with the same honest record.