Value Betting, Explained Without the Hype
A bet is not good because it wins. It is good because the price was wrong.
Value betting means backing a price that pays more than the true chance of the outcome deserves, not chasing whichever pick feels safest. Start from a fair reference: a deep, liquid exchange market, where the gap between the best back and lay prices is thin, gives a mid-price that reflects what a crowd of opposing bettors actually thinks the probability is. Compare that fair probability with a price quoted somewhere else. If the other price pays more than the fair probability implies, the expected value (EV) is positive, and a Kelly formula turns that edge into a stake, usually halved to smooth the ride. None of this proves the bet wins, and a single result never confirms or denies the probability either way.
Winning proves nothing. The price is the only evidence.
Ask a bettor who backed a 2.10 winner why it was a good bet, and most point at the result. That is not the whole answer. A coin flip called correctly is not evidence that the caller understands coins, and a single bet is a sample size of one. Ten bets is barely more. What actually separates a good bet from a lucky one is whether the price paid more than the true probability of the outcome deserved, and that question has nothing to do with what happened afterwards.
Every decimal price implies a probability: divide 1 by the price. A price of 2.30 implies roughly 43.5%. If the real chance of that outcome is higher than 43.5%, the price is generous. The catch is obvious once you say it out loud: what is the real chance?
One answer, imperfect but usable, sits inside a deep exchange market. Back and lay orders both sit on the same order book (backing and laying the same selection covers how those two sides actually work), and when a market is liquid the gap between the best back price and the best lay price narrows to a few ticks. The midpoint of that narrow gap is not a guess. It is the price a crowd of people, most of them staking real money against each other, currently treats as fair, which is a different kind of number to a bookmaker's price, built to balance a book and keep a margin rather than to reflect a consensus. Sharp money and how it prices differently is what actually narrows an exchange spread in the first place, and where a sharp reference price comes from covers the wider family of markets worth trusting for this.
The phrase itself is borrowed. Value investing meant buying a company for less than it was worth decades before anyone applied the same logic to a football match; the maths is different, the discipline of separating price from probability is not. Back to the price.
A market's depth matters as much as its number (a market minutes from kick off with three figure liquidity is not the same evidence as one with six figure liquidity an hour out). A tight spread on a thin book can look exactly like a tight spread on a deep one, right up until a real stake tests it.
The gap between two probabilities
Put a number on it. A bookmaker elsewhere quotes 2.30 for a selection. A deep, liquid exchange market for the exact same selection sits around 2.08, back 2.06 and lay 2.10, tight enough to trust the midpoint. Divide 1 by each price and the two implied probabilities sit apart.
A 4.5 percentage point gap is not proof either. It could mean the exchange has priced this correctly and the bookmaker is generous. It could also mean the exchange market is thinner than the spread suggests, or that something has moved since the mid-price was taken. Treat the gap as a hypothesis worth testing with a stake small enough to survive being wrong, not as a discovery.
The arithmetic, run once and checked twice
Same two prices, now turned into a stake: expected value first, then the two Kelly fractions, then cash on an assigned bankroll of £900.
| Step | Figure |
|---|---|
| Fair probability from the exchange mid-price (2.08; back 2.06, lay 2.10) | 48.0% |
| Bookmaker price for the same selection | 2.30 |
| Expected value: 0.48 × 2.30, minus 1 | +10.4% |
| Full Kelly stake fraction: 0.104 ÷ 1.30 | 8.00% |
| Half Kelly stake fraction | 4.00% |
| Full Kelly stake on a £900 bankroll | £72.00 |
| Half Kelly stake on a £900 bankroll | £36.00 |
Work it through in order. The exchange mid-price of 2.08 implies a fair probability near 1 divided by 2.08; call it 48% to keep the arithmetic clean, a working figure, not a fudge. The bookmaker's 2.30 carries no exchange commission (it is a straight bookmaker price, not an exchange leg), so the effective odds are simply 2.30. Expected value is probability multiplied by price, minus one: 0.48 times 2.30 is 1.104, minus 1 leaves +10.4%. That is the edge, expressed as a fraction of the stake, if the 48% estimate holds up.
Kelly turns that edge into a stake size rather than a flat amount. Full Kelly divides the edge by the price minus one: 0.104 divided by 1.30 is 8.00% of the bankroll. Half Kelly, the more common choice among people who size stakes for a living (see the arithmetic of staying profitable over time for why the smoother line usually wins over a long run), is simply 4.00%. On an assigned bankroll of £900, that is £72.00 staked at full Kelly, or £36.00 at half.
Turning that gap into a stake, automatically
Redo that arithmetic by hand for every price worth checking and the division becomes the reason most people quit before placing a bet. The calculator below does the same sum the moment a price and a probability go in, with an exchange commission option for legs that are not a plain bookmaker price like the one above.
It is turning that mispriced probability into a stake, and its own worked cases use different numbers again: type in 2.10 at a 55% estimate with no exchange commission ticked and it returns +15.5% EV and a half Kelly stake of 7.05%, worth £70.45 on a £1,000 bankroll. None of that, or the £72.00 above, proves anything on its own. The calculator sizes a stake for an edge you already believe you have. It does not go looking for one, and it cannot tell a correct 48% from a confident but wrong one.
Who this actually suits, and who it does not
Strengths
- Works from a price and a probability, not a hunch or a tip sheet.
- The stake shrinks on its own when the edge is small, instead of staying flat regardless of the gap.
Limits
- Needs a trustworthy fair price reference, which means checking a market's depth, not just reading its number.
- A run of losing "value" bets looks identical to a run of wrong probabilities; nothing here tells the two apart quickly.
Someone who already checks a few prices before staking and wants a disciplined way to size what they back.
Someone looking for picks, not a way to size stakes once picks are already chosen.
Arbitrage takes the opposite approach to a similar gap (the same idea used to lock a market instead): back every outcome at prices that already sum to a profit, and the result is locked in before the match starts, no probability estimate required at all. Value betting keeps the variance and asks for a better return over many bets instead of a certain one on this bet. Going further still, on a schedule that pays the bills, is its own separate decision; what going full time actually demands belongs to that decision, not to this page.
Reaching the market this page assumes you can already see
Everything above assumes access to a deep exchange market and, separately, to whichever venue is quoting the generous price. A direct Betfair account reaches the first for bettors it accepts. For everyone else, a broker's own exchange tool is the route in, and each one is a slightly different door (with its own exchange commission taken off before anything above applies).
MadMarket's Sharp Exchange states a flat 3% commission on winning bets, none on losing ones, and its terms turn away applicants from the UK, the USA and Australia among others. BetInAsia sells the same Sharp Exchange product; its current pages print no commission number (the last one found was 2.5%, so check the current terms), and its terms exclude the UK, the USA and France from registration. AsianConnect reaches the market through PIWIX or OrbitX, both said by AsianConnect's own help centre to carry a 3% exchange commission on winning bets ("may charge" is its own wording), so check the current rate; its own restricted country lists also disagree with each other, so check the current one rather than assume a country is covered. Sportmarket's FairExchange also charges 3% on winning bets only, and its terms name the UK, the USA and France among the countries it will not register.
Whichever door applies, tick the exchange commission box in the calculator above and use the real rate, not the example figure, before trusting a stake it returns. A broker's own screens can help with the price checking half of this too: the odds screens already sitting inside a broker account covers what each one actually shows before a bet goes anywhere.
If sizing a stake around a mispriced probability is not the right fit, it sits inside a wider set of approaches; the rest of the pro betting guides covers the others this site treats seriously, arbitrage and full time trading among them. Before funding any broker for this specific purpose, coverage and pricing side by side is worth reading in full, since the exchange commission and the country list both change which door is actually open.
Most bettors lose money over a long enough run, value betting included, once exchange commission, turnover rules and a moved price are counted in. Nothing on this page suggests this method, or any other, replaces an income. Bet only what you could afford to lose, and only from the legal age where you live.
Ask the price, not the outcome.