Market & prices
What value really means in horse racing
"Value" is the most-used and least-defined word in betting. It has a precise meaning — a price longer than the true chance deserves — and a problem nobody likes to mention: you never know the true chance. This guide covers both.
See where bookmakers beat the exchange today →What value means
A price is value when it is longer than the horse's true chance of winning deserves. In numbers: when the horse's probability of winning, multiplied by the decimal odds, comes to more than 1.
Value is about the price, not the horse. The most likely winner of a race can be poor value at a short price, and a horse that will probably lose can be good value at a long one. Value says nothing about whether this particular bet will win — only about what would happen if you could make the same bet at the same price many times.
The arithmetic, worked through
Worked example
You estimate a horse's chance of winning at 25%. Your fair odds are 1 ÷ 0.25 = 4.00 (3/1).
The best price on offer is 5.00 (4/1). That price implies a 20% chance.
Expected return per £1 staked = probability × decimal odds − 1 = 0.25 × 5.00 − 1 = +£0.25.
If — and only if — your 25% is right, backing horses like this at 5.00 would return 25p profit per £1 on average over a long run. Three bets in four would still lose.
That "if" carries all the weight. The price is known exactly. Your probability is an estimate.
Model error: the part nobody quotes
Any probability — yours, a tipster's or a model's — has error in it. The table shows what happens to the same 5.00 bet if the true chance is a little different from the 25% estimate:
| True chance of winning | Expected return per £1 |
|---|---|
| 18% | −£0.10 |
| 20% | £0.00 |
| 22% | +£0.10 |
| 25% (the estimate) | +£0.25 |
| 28% | +£0.40 |
An estimate five percentage points too high turns an apparent 25% edge into nothing. Estimating a horse's chance to within five points is hard; doing it consistently across hundreds of races is harder. And on an exchange, commission moves the break-even point further: at 5.00 with 5% commission on winnings, you need a chance of about 20.8% just to break even.
Why the market is hard to beat
The price you are comparing against is not a guess. It is the combined judgement of everyone betting on the race — including people with better information than any public data source, and professional bettors whose models are built for exactly this job. Over large samples, racing markets are well calibrated: horses priced at around 20% win around 20% of the time, give or take the margin and some well-known biases at the extremes.
So when your estimate and the market disagree, the likelier explanation is usually that your estimate is wrong. Finding genuine value means having a reason to believe you know something the price does not — and then checking, over a large sample, that you did.
Our own result. We back-test every scoring model we run against real results. Our honest finding is that our model ranks a race sensibly but tracks the market rather than beating it: its top picks do not turn a profit at the prices you can actually get. We publish that, with the full record, on the performance page — which is why we never present a score or a pick as value, or as a reason to bet.
Even real value loses — often
Suppose you really did have a 10% edge: a true chance of 22% at 5.00, bet after bet. Over 200 such bets you would still be behind roughly one time in five. Over 1,000 bets, about one time in seventeen. Winning runs and losing runs from luck alone are long enough to fool anyone who judges a method on a few weeks of results — in either direction.
Two different things called "value"
- Value against your own estimate. The price is longer than your probability says it should be. Only as good as the estimate.
- Value against another market. A bookmaker offers a longer price than the margin-free exchange price. This is a comparison between two sets of prices, and it assumes the exchange price is the better estimate of the horse's chance — often a reasonable assumption on busy markets, not a proof.
The second is simpler and needs no view on the horse at all. It still carries risk, and whether it pays over time depends on prices you can actually get, in the stakes you actually bet.
Common mistakes
- "Value" as a synonym for "a horse I like at a big price." Without a probability estimate, there is no value calculation — just an opinion and a price.
- Judging value by results. A value bet can lose and a bad bet can win. Only a long record, settled at the prices you took, tells you anything.
- Ignoring the margin and commission. They are the first few percentage points any edge has to clear.
- False precision. A 25% estimate is really "somewhere around 20–30%". Treat small apparent edges accordingly.
What TRF shows
The Value board lists runners where a bookmaker is offering a longer price than the margin-free Betfair Exchange price — value in the second sense above. It is arithmetic between two markets, with no prediction model involved, and it is published unvalidated: we will not claim it is profitable until there is a large enough sample to say.
TRF does not publish its own probabilities or tell you what is value. Our AI Score is a ranking of the runners in a race, not a probability; our performance page shows how it compares with the market, losing runs included. The fair odds guide covers the probability arithmetic behind all of this.
Put it into practice
The Value board shows where a bookmaker is offering a longer price than the margin-free exchange price. It is arithmetic between two markets, published unvalidated.
See where bookmakers beat the exchange today →The Racing Formula is a free racing research site. Nothing here is betting advice, and no statistic, score or price on this site is a guarantee of anything. 18+. If gambling stops being fun, BeGambleAware and GamStop can help.