Closing Line Value: Calculation, Uses and Limits

Define the close before calculating it

A closing record needs source, market, selection, settlement period, timestamp rule, price type, availability and treatment of suspended or missing markets. Betfair's historical data specification documents timestamped prices, traded volume, best available levels and market status, illustrating why “the closing price” is not one self-defining field.

Possible benchmarks include the last available sportsbook quote before suspension, a named exchange price at a fixed pre-kickoff offset, or a composite of named sources. Choose one before seeing results and do not switch to whichever close makes the record look strongest.

Two transparent calculations

Suppose the accepted decimal price was 2.20 and the defined closing price was 2.00.

Price-ratio CLV = (2.20 / 2.00) - 1 = 0.10, or 10%.

Raw break-even probabilities give another view:

• Accepted raw probability = 1 / 2.20 = 0.45455, or 45.455%.
• Closing raw probability = 1 / 2.00 = 0.50, or 50%.
• Raw probability move = 50% - 45.455% = 4.545 percentage points.

These measures answer different questions. Report the formula and avoid calling a 10% price ratio a ten-percentage-point probability gain.

Remove margin consistently

For multi-outcome fixed-odds markets, reciprocal prices contain overround. A fairer probability comparison needs every outcome at both timestamps and one declared margin-removal method. Research shows that source and conversion method affect odds-derived forecasts (Štrumbelj, 2014). Do not de-margin the opening quote with normalization and the close with a different method.

What CLV can and cannot show

Historical research found forecast-quality differences among bookmakers and leagues (Štrumbelj and Robnik-Šikonja, 2010). That supports recording the benchmark identity; it does not establish a permanent ranking of current operators.

Minimum CLV report

• Number of qualifying decisions and coverage rate.
• Median and mean price-ratio CLV.
• De-margined probability-point movement where complete books exist.
• Distribution, not only the average.
• Results by market, league, lead time and source.
• Missing, suspended, rejected and partially accepted prices.
• Comparison with realised returns and model calibration, reported separately.

Calibration guidance remains necessary because favourable price movement does not validate the absolute model probabilities.

Test benchmark sensitivity

Calculate CLV against at least two predeclared alternatives where data permits, such as a named sportsbook close and an exchange observation five minutes before scheduled kickoff. Keep missing rows missing; do not substitute one benchmark only when it is favourable.

Report how many decisions change sign under the alternative close, whether the result depends on low-volume markets, and whether source coverage is balanced across leagues and price bands. The market-source differences documented by Štrumbelj make this sensitivity test substantive rather than cosmetic. A conclusion that survives both benchmarks is more informative, but still remains benchmark-specific.

Continue learning

• Next guide: Lower-League Value Betting
• Related guide: Using Betting Exchanges as a Probability Benchmark
• Definition: Closing Line
• Definition: Closing Line Value

Assumptions and limitations

The example is illustrative and ignores commission. CLV is a later market comparison and may be useful as a diagnostic, but it is not labelled the best or gold-standard measure of skill. The concise CLV glossary definition covers terminology; this page owns benchmark design and reporting.