Closing Line Value (CLV) Explained
Closing Line Value (CLV) is the gap between the price at the moment an estimate was made and the market's closing price: if your estimates consistently stand ahead of the close, your process is beating the market's most informed state. Short-run hit streaks are mostly luck; systematic positive CLV is the most reliable known indicator of forecasting skill. 11Stat tracks CLV as a model-quality metric and reports it openly — this page is an analytical explainer, not betting advice.
The definition: where did you stand relative to the close?
Suppose an assessment made on Monday corresponded to a 50% price, and by kickoff the market had priced the same outcome at 55%. The close moved in your direction: that is positive CLV. Had the close come in at 45%, it would be negative CLV.
The power of CLV is that it ignores the single match result. Whatever happens that night, it measures whether you stood ahead of or behind the market's final, fully informed state. Of all the concepts in the glossary, it is the most purely process-oriented one.
Why the closing line is the benchmark
The closing price is the final form of the market after everything that flowed in between opening and kickoff — lineups, injuries, weather, money — has been priced. The repeated finding of the empirical literature is that the closing price is the most accurate known single estimate of match probabilities: it errs less than the opening price and less than nearly every individual model.
That is market efficiency in practice: the knowledge of thousands of participants aggregates into one price. This is what makes "beating the close" a demanding bar — and precisely why it is an honest quality measure. The underlying price-probability mechanics are covered on the odds vs model probability page.
CLV vs short-run results: process beats scoreboard
Football is a high-variance sport: a poor process can look brilliant for a few weeks, and a good process can run cold for months. A ten-match hit streak — or losing streak — says almost nothing statistically; the sample is far too small.
CLV shortens that noise. Because it measures a price comparison instead of an outcome, every single estimate yields a meaningful data point, and skill separates from luck at a much smaller sample size. It is the numerical form of the process-versus-results distinction: in the short run you read where the process stands against the close, not the scoreboard. Together with calibration, CLV is one of a model's two fundamental honesty tests.
How CLV is computed: a concrete example
The common computation compares the margin-stripped (no-vig) closing probability with the probability at assessment time. Example: no-vig probability 48% at assessment, 53% at the close → CLV = +5 points (or +10.4% proportionally). Across hundreds of estimates, the average CLV and the share of positive-CLV estimates are reported.
Details that matter: the comparison must use the same market and the same de-margining method; the full distribution should be read, not a few large positives; and in small samples even CLV is noisy — meaningful reading takes hundreds of data points.
How 11Stat uses CLV
On 11Stat, CLV is a model-quality metric, not a call to action. In walk-forward (leakage-free) backtests, every model estimate is compared against that match's no-vig closing probability; average CLV, the positive-CLV share and per-market breakdowns are published alongside the simulated ROI reports. When paper ROI swings while CLV stays steady, that steadiness is the least misleading indicator of whether the model produces real signal.
The measurement chain — methodology, calibration, CLV — is mutually auditing: methodology shows how the numbers are produced, calibration shows whether the numbers can be trusted, and CLV shows where the process stands against the market's most informed state.
The limits of CLV: it is not a certainty either
CLV is powerful but not flawless. In thin, lightly traded markets the closing price is less informed and the comparison weakens. The choice of de-margining method can shift results by a few points. And like every metric, it misleads at small samples: a few positive weeks are not evidence of skill.
Most importantly: positive CLV indicates a healthy process; it does not promise any outcome, and 11Stat never presents it as if it did. CLV is an analytical evaluation tool — not an instruction to act.
Frequently Asked Questions
Does positive CLV mean profit is guaranteed?
No, it does not. Positive CLV only shows that assessments stood ahead of the market's most informed final state; individual results remain subject to high variance, and even long positive-CLV stretches can overlap with losing runs. CLV is a process indicator, not a promise of outcomes.
Is CLV a betting strategy?
No; in 11Stat's context CLV is a model-quality diagnostic. It measures and reports where the model's estimates stood relative to the closing price — it contains no recommendation to act. 11Stat gives no betting advice.
Is consistently beating the close a sure thing for making money?
No — a sure thing does not exist in a probabilistic domain. Systematic positive CLV is strong evidence that a process produces information, but whether that converts to returns depends on margins, variance and sample size, and it never becomes certain. On 11Stat all such measurements are paper-simulation metrics.
Why look at CLV instead of a short-run hit rate?
Because short-run hit rates are largely a function of luck: ten-match streaks prove close to nothing. CLV converts every estimate into an outcome-independent price comparison, so skill separates from luck at a much smaller sample. If the process is good, CLV shows it long before the scoreboard does.
How large a sample does CLV need?
There is no exact threshold, but it takes hundreds of estimates rather than dozens — more still if you break results down by market and league. That is why 11Stat publishes CLV in periodic backtest reports in aggregate, not as commentary on individual matches.
Does 11Stat measure CLV with real money?
No. All CLV and ROI measurements on 11Stat are leakage-free walk-forward simulations over historical data — paper metrics. Their purpose is auditing model quality; there is no real-money flow, slip or betting transaction anywhere on the platform.