Closing line value, usually abbreviated to CLV, compares a recorded earlier price with a closing-market reference for the same selection. It helps separate the price you obtained from whether one match happened to win. There is more than one calculation called CLV, so a useful report states its formula and reference source.
Define the closing reference first
Choose the bookmaker, event, market period, line and final pre-match observation rule before looking at results. For an archive, “closing” normally means the last captured quote before the provider's cut-off. It is not necessarily the last price any user could have obtained. A record taken after kick-off cannot serve as a pre-match close.
The OddsTips guide to odds movement explains why price observations need consistent source and timing definitions. The same requirement applies to a CLV log. Missing closing quotes should remain marked as missing under a rule you chose in advance.
Calculate a raw price-ratio measure
Suppose you record decimal odds of 2.20 for a selection and the equivalent closing quote is 2.00. One simple definition is raw CLV = (earlier odds ÷ closing odds − 1) × 100. Here that gives (2.20 ÷ 2.00 − 1) × 100 = 10%.
Under this definition, your earlier quote offers 10% more gross return than the closing quote for the same winning selection. It does not mean your realised profit is 10%, your chance of winning improved by ten percentage points, or your estimated edge is automatically 10%.
The raw implied probabilities are 45.45% at 2.20 and 50% at 2.00. That is a 4.55 percentage-point difference. A chart showing probability-point changes must be labelled differently from a chart showing price ratios.
Remove margin for a probability-based benchmark
Now suppose the complete two-outcome closing market is 2.00 and 1.90, with no draw or push. Its raw probabilities are 50% and 52.6316%, totalling 102.6316%. With proportional normalisation, the first selection's closing fair-probability estimate is 0.50 ÷ 1.026316 = 48.7179%.
The corresponding fair closing price is approximately 2.0526. Against that reference, the earlier 2.20 quote has an estimated return of 2.20 × 0.487179 − 1 = 7.18% per unit staked. Equivalently, 2.20 ÷ 2.0526 − 1 gives the same result, allowing for rounding.
The raw measure was 10%; the margin-adjusted estimate is 7.18%. Neither number is a measured profit. The second calculation assumes the chosen margin-removal method produces a useful probability estimate. Our fair odds article explains that assumption and alternatives.
Keep the result in another column
If this one-unit selection wins at 2.20, its net result is +1.20 units. If it loses, the result is −1 unit. Neither outcome changes the recorded CLV. A positive-CLV selection can lose, and a negative-CLV selection can win.
Over a sample, report the number of qualifying records, missing closes, the CLV distribution and realised results separately. If stakes vary, distinguish a simple average across selections from a stake-weighted average. Otherwise one report can quietly give large and small positions the same importance.
Know where the benchmark can fail
A stale price, a lightly covered market or mismatched settlement rules can make the comparison misleading. A closing reference may contain useful later information without being a perfect estimate of the outcome probability. The quality of the benchmark needs checking across the sport and market you actually study.
Changing a handicap or total line also changes the selection. An earlier over 2.5 price cannot be divided by a closing over 3.0 price and interpreted as ordinary same-selection CLV. Pushes, quarter lines and exchange commission require calculations appropriate to their payoff rules.
Use CLV as one diagnostic in a documented research process. Pair it with consistent price comparisons, testing on later data and a clear record of assumptions. It is evidence about pricing, not a guarantee of profitable future results.