The short answer
Positive expected value is mathematically possible, but a label such as “value betting” does not establish that the underlying probabilities are accurate or that quoted prices can be executed. The evidence required in 2026 is the same core evidence required in any period: timestamped forecasts and prices, later-sample validation, complete settlement, costs, and uncertainty.
Academic football findings are not uniform. A 33,060-match historical study found different efficiency results by league and by mean versus best available prices (Angelini and De Angelis, 2019). A later study found different probability-bias conclusions for Premier League and La Liga data (Koning and Zijm, 2023). These studies show why a current claim needs its own evidence; they do not supply a 2026 expected return.
Evidence ladder for a profitability claim
| Level | Evidence | What it can support |
|---|---|---|
| 1 | Hand-picked winning examples | Illustration only |
| 2 | Settled record without pre-event timestamps | Historical summary with selection-bias risk |
| 3 | Preserved forecasts and offered prices | Recalculation of stated decisions |
| 4 | Accepted prices, stakes, costs and all qualifying events | Implemented historical result |
| 5 | Locked method evaluated on a later untouched period | Stronger prospective evidence for that period |
| 6 | Repeated later-period results with calibration and drift monitoring | Maintained evidence, still not a guarantee |
Recalculate the headline
For 800 one-unit settled bets, 760 units returned, 20 units voided and refunded, and 20 units lost:
- Total stake at risk in the recorded convention = 800 units.
- Net profit = 780 - 800 = -20 units when refunds are included in total returns.
- Yield = -20 / 800 = -0.025, or -2.5%.
The reporting convention must state whether void stakes are included in turnover and returns. Do not mix gross return, profit, ROI, yield and strike rate.
Current execution checks
Exchange commission may be charged on net market winnings and can vary by account and market; Betfair's current commission documentation is one first-party example. Fixed-odds constraints, rejected bets, partial acceptance and price changes also need recording. A backtest at prices that were never available for the required stake does not demonstrate implementation.
Questions a credible record answers
- Were forecasts published or hashed before events?
- Was the selection rule fixed before the evaluation period?
- Are all eligible bets and passes present?
- Are accepted prices used rather than later closing or best-screen prices?
- Are model probabilities calibrated? Calibration guidance explains the frequency check.
- Are uncertainty intervals and drawdown reported?
- Have rule changes started a new evaluation series?
Practical conclusion
Do not ask whether value betting in general is profitable. Ask whether one precisely specified process produced reliable probabilities, executable positive-EV decisions and complete later-period results after costs. Until those records exist, “profitable” is an unverified claim.
Next step
Use Market Efficiency Betting for the next part of this topic.
Continue learning
- Next guide: World Cup 2026 Dark Horses
- Related guide: Odds Drift
Assumptions and limitations
This page is current to 14 July 2026 and contains no current operator ranking, live odds or performance forecast. It should be reviewed by 12 October 2026. Betting can cause financial and mental harm; the NHS guide provides support and practical safeguards.
GambleAware
GamCare
Gambling Therapy
GAMSTOP