Football Betting-Market Efficiency: How to Read the Evidence

Define the claim precisely

“The market is efficient” is incomplete. A test must name the information set, competition, market, operator or exchange, price timestamp, margin method, sample period, execution rule and costs. The Angelini and De Angelis study defines and tests weak-form efficiency through information in football prices and forecast errors; broader information sets answer different questions.

What football research actually finds

The Angelini and De Angelis study found eight efficient and three apparently inefficient markets when using best prices in its design. Koning and Zijm found different residual probability biases in their Premier League and La Liga applications. These results argue against both “all markets are unbeatable” and “niche markets always contain easy value.”

Forecast efficiency and betting profitability differ

A price can be a strong probability forecast yet still be unprofitable to back after margin. Conversely, a model may improve a probability score slightly without creating an executable price advantage. Goddard's football study evaluates forecast and betting results as related but distinct evidence, motivating these three layers:

1. Forecast quality: proper scores, calibration and uncertainty.
2. Price disagreement: model probability versus timestamped executable price.
3. Realised implementation: accepted stakes, costs, settlement and returns.

Favourite-longshot bias is not one fixed adjustment

Favourite-longshot bias describes a systematic relationship between price level and observed frequency or return. The direction and size can differ by outcome type, league and method. Koning and Zijm show why basic normalization may leave a structured bias and why even Shin probabilities require empirical checking.

Design a prospective test

1. Declare the hypothesis and selection rule.
2. Build the rule using an earlier development period.
3. Lock model, price cutoff, de-margin method and staking unit.
4. Apply it to an untouched later period.
5. Record every qualifying event and execution failure.
6. Report probability scores, calibration, return, uncertainty, drawdown and coverage.
7. Repeat after enough new data without silently changing the rule.

Avoid searching many leagues, thresholds and methods and reporting only the strongest combination. That converts noise into a story unless multiplicity and an untouched confirmation period are handled.

Next step

Use Is Value Betting Profitable 2026 for the next part of this topic.

Continue learning

• Next guide: How Bookmakers Set Football Odds
• Related guide: Kelly Criterion for Value Bettors

Assumptions and limitations

The cited papers are evidence about their own historical designs, not current operator rankings or guaranteed strategies. Market structure, margins, limits, data and participants can change. This page teaches how to interpret evidence; it does not state that any current football market is efficient or inefficient.