Value betting and expected value, explained properly

What a fair price is

A value bet is a bet where the odds on offer are higher than the real chance of the outcome justifies. Unlike a sure bet, you back only one side. So any single value bet can lose, and many of them will. The idea only works as an average over a large number of bets, and even then results swing a lot.

The hard part is knowing the real chance. Nobody knows it exactly, but some prices are much better estimates than others. Low-margin "sharp" bookmakers and busy betting exchanges take large bets from well-informed players and move their prices quickly. Their odds, especially shortly before the start, are the best public estimate of the true probability.

A reference is only as good as the market behind it. A top-league football match an hour before kick-off has a very efficient sharp price. A lower-league game two days out does not. Be more sceptical of an edge found against a thin or early price, and ask for a bigger margin before you bet it.

So the standard method is: take the sharp price, strip out its margin to get a fair price, and look for other bookmakers offering more than that.

Removing the margin

A bookmaker's implied probabilities (1 ÷ odds) add up to more than 100%. The excess is its margin, also called the overround or vig. To get fair probabilities you take it out.

The multiplicative method is the simplest. Divide each implied probability by the total:

fair p = (1/odds) / Σ(1/odds)
fair odds = 1 / fair p

The power method instead finds an exponent k so that Σ(1/odds)^k = 1. It takes slightly more margin off outsiders than favourites, which matches how bookmakers tend to price. For two near-even outcomes both methods give almost the same answer; for a 1.20 favourite against an 8.00 outsider the difference matters.

Expected value

Once you have a fair probability p, the expected value of a bet at decimal odds is:

EV = p × odds − 1

The result is the average profit per euro staked. An EV of 0.05 means €5 per €100 on average, over a very large number of similar bets. It says nothing about the next bet.

Real edges are usually smaller, around 2% to 5%. Anything far above that deserves a second look: the sharp price may have moved, it may be a different market (sets vs games, extra time included or not), or the price may be an error the bookmaker later voids.

Variance: why 100 bets prove almost nothing

This is the part most people underestimate. Take flat €10 stakes at odds of 2.00 with a 4% edge, which means a true win chance of 52%. Each bet ends at +€10 or −€10, and the spread of results over n bets is roughly:

SD of profit ≈ stake × odds × √(n × p × (1 − p))

After 100 bets the noise is two and a half times the signal. After 1,000 bets, someone with a genuine 4% edge still has roughly a 1 in 10 chance of being behind. Only over several thousand bets does the edge reliably show. Higher odds make the swings larger, and if your fair prices are off, the edge may not exist at all.

In practice that means two things: a bankroll big enough to sit through losing runs (see the bankroll guide), and a better way to judge your process than last month's profit.

Flat stakes and fractional Kelly

The simplest approach is flat staking: the same stake, one "unit", on every value bet, for example 1% of your value-betting bankroll. It is easy to track and hard to get badly wrong.

The Kelly criterion instead sizes each bet by its edge, to maximise long-run growth:

Kelly fraction = (p × odds − 1) / (odds − 1)

In the example above: 0.075 ÷ 1.15 = 6.52% of the bankroll. That assumes your fair probability is exactly right, which it never is. Full Kelly with an overestimated edge leads to deep drawdowns, so most people who use it bet a quarter or less (here about 1.6%) and cap any single stake at 2% to 3% of the bankroll. Recalculate from your current bankroll, so stakes shrink in a losing run.

Closing line value: better evidence than profit

Because results are so noisy, it helps to track closing line value. Compare the odds you took with the fair sharp price just before the start:

CLV = your odds / closing fair odds − 1

If you took 2.15 and the fair closing price was 2.05, your CLV is +4.9%. The closing price contains everything the market knew, so consistently beating it is a good sign that your prices really were above fair, often visible within a few hundred bets, long before profit settles down. It is still evidence, not a promise of future results.

Beating the closing price is also exactly what bookmakers watch for. Expect stake limits sooner or later; read how bookmaker limits work.

Value bets in Austria and Germany

Two local details change the maths:

• German betting tax. In Germany a 5.3% tax applies to stakes. Some bookmakers pay it themselves; others pass it on to you. If it is passed on, it cuts roughly five percentage points off your edge, which turns a typical 3% to 4% value bet negative. Always check the net price.
• Licensed bookmakers. In Germany you may only bet with operators on the official licence list (GGL whitelist). In Austria the choice is wider, but read each bookmaker's rules on voids, settlement and stake limits before you rely on its prices. Winnings from private betting are generally tax-free in both countries; details are in tax and legality.

Sure betting vs value betting

Many people combine both, using sure bets for steady turnover and a small, separate bankroll for value bets. Either way, losing days are part of betting. Only stake money you can do without, and if betting stops being fun, our responsible gambling page lists free help.

Not betting advice. 18+. Results vary and can be negative.