What Is the Closing Line?
A closing line is the benchmark price captured at a stated time near the start of an event. It is not a single universal number: operators and exchanges may show different prices, margins, liquidity, and settlement rules. Define the source and timestamp before collecting data.
Closing prices are useful because they are observable and later than the decision-time price. Research has tested the forecasting ability and efficiency of fixed-odds football markets, but findings are tied to the competitions, periods, prices, and methods studied rather than establishing that every closing price is true probability. See Goddard and Asimakopoulos' football forecasting study for one peer-reviewed example and its sample-specific design.
How CLV Works
If you place a bet at odds of 2.50 on Arsenal to beat Newcastle, and the closing odds for that same selection drop to 2.20, you have beaten the closing line. Your odds were better than the market's final assessment.
Conversely, if you accept 2.50 and the comparable close is 2.80, the later market offered a longer price. That is negative CLV under this convention; it does not by itself reveal which probability was correct.
One simple decimal-price comparison is:
Price CLV = (accepted decimal odds / closing decimal odds - 1) x 100
Using the first example: (2.50 / 2.20 - 1) x 100 = 13.6%. Label this as a price ratio, because it is not the same as a 13.6 percentage-point probability edge. For cross-market analysis, compare de-margined probabilities and account for exchange commission.
What CLV Can and Cannot Show
CLV can show whether accepted prices systematically differ from a later benchmark. This is useful for auditing timing, price capture, and model decisions without grading every decision solely by a noisy match result. It should accompany direct forecast evaluation such as the probability checks in scikit-learn's calibration guidance.
CLV cannot show that the close was perfectly calibrated, that the original model was well specified, or that an apparent pattern will survive fees and future data. A positive average can also be distorted by a few large ratios, stale or untradeable closing quotes, inconsistent sources, or selecting the benchmark after seeing the result. Probability forecasts should also be tested directly for calibration; scikit-learn's calibration guide explains the relationship between predicted probabilities and observed frequencies.
A Football Example
Assume an illustrative selection is accepted at 2.30 and the pre-defined closing source later records 2.05 for the same market and settlement rules.
The price CLV is (2.30 / 2.05 - 1) x 100 = 12.2%. The two raw implied probabilities are 1 / 2.30 = 43.5% and 1 / 2.05 = 48.8% before removing market margin. The movement is consistent with the entry price being better than this close, but it does not confirm the model's probability; calibration must be assessed separately against observed outcomes as described in scikit-learn's guide.
The match result does not change the recorded price comparison. Equally, positive CLV on one bet does not validate the process; it is one observation in an evaluation that also needs forecast quality, execution costs, and uncertainty.
How to Track Your CLV
Tracking CLV requires a consistent recording process:
- Log the odds at which you place every bet
- Pre-define the close by source, timestamp, market, and price availability
- Check comparability of settlement rules, margin, commission, and currency
- Calculate consistently in price or de-margined probability space
- Review the distribution by market and time, with uncertainty and outliers visible, alongside direct probability calibration
A spreadsheet can automate the arithmetic, but it cannot repair inconsistent inputs. Preserve both raw prices and the transformation so another reader can reproduce the measure.
Limitations of CLV
Closing quotes may be stale, limited, suspended, untradeable at the recorded stake, or drawn from a differently margined market. Market selection can also bias the metric if only favourable movements are retained.
CLV and realised return answer different questions. Either can be positive while the other is negative over a finite sample. Report both, avoid a universal sample-size rule, and do not use CLV as a substitute for direct probability validation.
Past performance does not guarantee future results. CLV is a useful analytical tool, but it does not eliminate the risk inherent in all forms of betting.
18+. Gambling involves risk. For support, visit begambleaware.org. Please gamble responsibly.
Related resources
Continue with Value Bet: What It Means in Betting for the next part of this topic, or return to Betting Glossary: Every Betting Term Explained in Plain English to compare the other guides in this collection.
Continue learning
- Next guide: Correct Score Betting
- Related guide: Correlation in Betting
- Go deeper: Closing Line Value
Assumptions and limitations
CLV depends on a closing benchmark chosen before results are inspected. Source, timestamp, market rules, margin treatment, liquidity, and available stake must remain comparable. The cited Goddard and Asimakopoulos study is sample-specific evidence about football forecasting and market efficiency; it does not prove that every close is a true probability or that positive CLV guarantees profit.
GambleAware
GamCare
Gambling Therapy
GAMSTOP