Value Betting Basics: The Long-Run Math
Value betting is often described as backing the winner at a bigger price than the true probability deserves. That definition is correct but incomplete. Value is a relationship between price and probability, and it says nothing about whether the bet wins. A value bet can lose ten times in a row and still be the right bet, which is why the long run is the only place the maths shows up.
Most bettors abandon a sound method during the first losing run, and most of the rest abandon it during the first winning run, when they increase stakes faster than the edge justifies. The mathematics below is not complicated, but it is the difference between a hobby and a process that survives a season.
What value actually means
Expected value is the average result of a bet repeated many times. For a bet at decimal odds of 2.50 with a true probability of 45%, the expected value per unit staked is 0.45 x 1.50 minus 0.55 = 0.125, or 12.5%. The bet is profitable in expectation even though it loses more often than it wins.
A useful sanity check is to convert your estimate back into a price. If you believe a team has a 40% chance, the fair price is 2.50, and anything above that is value. If the market offers 2.30, the same opinion is a pass, no matter how confident you feel about the team. The discipline lies in applying the same conversion to every bet, including the ones you like.
The true probability is the hard part. It comes from your model, your records or your judgement, and it is never known exactly. A 3% edge estimated from 40 bets is mostly noise. A 3% edge measured over 2,000 bets is a signal worth acting on.
Why the best team loses
Sport is a low-scoring, high-variance process. A team that creates 2.1 expected goals and concedes 0.8 wins perhaps 65% of the time, not 100%. The same team can lose to an own goal and a deflection. If you bet on it at fair odds, you will still lose 35% of the time, and the sequence of losses can be long enough to make a good bettor doubt the method.
This is the psychological core of value betting. The result of a single bet tells you almost nothing about the quality of the decision. Only the price and the estimated probability tell you that, and only a large sample tells you whether your estimates were right.
How large a sample you need
A rough formula for the number of bets needed to be 95% confident of showing a profit is 1.96 divided by the edge, squared. At even money, a 3% edge needs about 4,300 bets and a 5% edge needs about 1,540. The numbers fall quickly as the edge rises, which is why specialists with a narrow niche do better than generalists with a small edge everywhere.
There is a shortcut worth knowing. If you track closing line value, you can estimate your edge before the results arrive, because the closing price is the market's best estimate. Beating the close by 2% on 200 bets is far more informative than a 15% yield on 40 bets, and it arrives much sooner.
Variance also depends on the odds. A 10% edge on a 1.20 favourite produces frequent small wins and a smooth curve. The same edge on a 6.00 outsider produces long losing runs and a much wider range of outcomes, even though the expected value is identical.
Edge, distance and drawdown
The table converts an edge into the sample size and the drawdown you should expect to survive. The drawdown column assumes flat one-unit stakes at even-money odds, and it is a planning range rather than a worst case.
| Edge | Bets for 95% confidence | Expected worst drawdown |
|---|---|---|
| 2% | 9,600 | 150 to 250 units |
| 3% | 4,270 | 90 to 160 units |
| 5% | 1,540 | 50 to 90 units |
| 7% | 785 | 30 to 60 units |
| 10% | 385 | 20 to 40 units |
| 15% | 170 | 12 to 25 units |
Staking and survival
The fastest way to turn a profitable method into a losing account is to stake too much during a drawdown. Flat staking keeps the stake constant and makes the maths easy to audit. Percentage staking reduces the stake as the bank shrinks, which lowers the risk of ruin but also slows recovery. Our comparison of flat and percentage staking covers the trade-offs in detail.
Keep a reserve outside the betting bank. The reserve is not for chasing losses; it is there so that a normal drawdown does not become an existential one. A bank of 100 units with a 50-unit reserve behaves very differently from a bank of 100 units and no reserve, even at identical stakes.
The practical rule is to size bets so that a 100-unit drawdown is uncomfortable but survivable. If a 20-unit losing run would force you to stop, your stakes are too large for the edge you have.
Tracking the evidence
You cannot manage what you do not record. A simple sheet with the date, selection, price taken, closing price, stake and result is enough to calculate yield, average odds and closing line value. The closing line is the most useful single number, because beating it consistently is evidence of skill even when short-term results are poor. A ready-made structure is described in our guide to tracking bets in a spreadsheet.
- Record the price you took at the moment you placed the bet.
- Record the closing price, not the price an hour before kickoff.
- Group results by market, because edges are market-specific.
- Review the sample at 500 bets, not at 50.
- Change the method only when the data, not the mood, says so.
Value betting is a business of small edges repeated many times. The maths is simple, the discipline is not. Estimate the probability, take the price only when it is wrong, size the stake so you survive the variance, and judge yourself on the process rather than on the last result.
