English 
Español 
Português 
Deutsch 
Français The counterintuitive core
The most important idea in betting is that picking winners and making money are different skills. A bet on a 90% favourite at odds of 1.05 wins nine times out of ten and still loses money, because the price pays less than the risk deserves. A bet on a 30% underdog at 4.00 loses seven times out of ten and profits, because the price pays more.
Value is the gap between the odds on offer and the outcome's true probability. Betting is profitable exactly when you find prices that underestimate real chances — and only then. Everything else is entertainment with a fee.
Expected value in one formula
Expected value (EV) per unit staked: (probability × odds) − 1. If you estimate a team's chance at 45% and the price is 2.40, EV = 0.45 × 2.40 − 1 = +0.08 — eight cents of long-run profit per euro staked. At 2.10 the same estimate gives −0.055: a losing bet regardless of how the match ends.
The formula is trivial; the probability estimate is everything. Your 45% has to be closer to the truth than the market's implied 41.7% (at 2.40, margin removed) for the edge to be real. That is a high bar against modern football markets.
Where value hides
Big markets like Premier League match odds are efficient — thousands of sharp participants push prices close to true. Value hunters look where attention is thinner: smaller leagues, niche markets, early lines before the market settles, and price discrepancies between bookmakers.
Line shopping is the one source of value available to everyone with zero modelling: the best available price across bookmakers is routinely 3–8% better than the average price. Taking 2.02 instead of 1.90 on the same outcome needs no forecast at all, just an extra tab.
How our screener applies this
Our model estimates probabilities from historical scoring rates (a Poisson framework blended with market prices), converts them to fair odds and compares against the best available price. When the price exceeds the model's fair odds by enough, the pick gets a positive edge score — that is the number in the Edge column of the screener.
The honest caveats: model probabilities are estimates with error bars, edges of a few percent are easily illusions, and even genuine edges lose often — value betting profits emerge over hundreds of bets, never within a weekend. Our published closing-line-value tracking exists precisely to test whether the model's edges are real.
You estimate a draw at 30% (fair odds 3.33). Bookmaker A prices it 3.10 — negative EV, pass. Bookmaker B prices it 3.60: EV = 0.30 × 3.60 − 1 = +0.08. Same match, same estimate, one bad bet and one good one. The market decided which — not the final score.
- Value = price above true probability; EV = (prob × odds) − 1.
- Likely winners at short prices can be terrible bets; unlikely winners at long prices can be great ones.
- Line shopping across bookmakers is free value requiring no forecasting skill.
- Real edges are small, noisy and only visible over hundreds of bets.
Frequently asked questions
How do I know if my probability estimates are any good?
Track them against closing odds: if you consistently beat the closing line, your estimates carry information. That signal shows up within weeks, long before profit curves become meaningful.
Is value betting the same as arbitrage?
No. Arbitrage locks in a small guaranteed profit by covering all outcomes across bookmakers; value betting takes on variance, backing one outcome at a price it believes is wrong. Value betting can lose for months even when done well.
Can I value bet with a small bankroll?
The maths works at any size, but variance is brutal: even with a real 5% edge, hundred-bet losing stretches of the bankroll happen. Flat staking and a genuine entertainment budget are prerequisites, not options.