Closing line value: the only honest test of a prediction model
A model can go 6-4 and be excellent, or 7-3 and be worthless. The difference shows up in whether it beat the closing price.
Closing line value measures whether the price you took was better than the price available just before kick-off. Beating the close consistently implies your estimate was sharper than the market's final consensus, independent of whether the bet won.
Why win rate is a bad scoreboard
Short-run results are dominated by variance, so a win-rate table mostly ranks luck.
At odds around 2.00, a genuinely skilled bettor and a coin-flipper can produce identical ten-bet records. Over a season the noise shrinks, but by then you have spent a year and learned very little about whether the method works.
Worse, win rate rewards overconfidence. A model that only bets heavy favourites will show a flattering strike rate while losing money to the margin, because the prices it accepts are too short.
How closing line value isolates skill
CLV compares your entry price to the closing price, which is the market's most informed estimate.
The closing price is the most information-dense number in the market: it incorporates every bet placed, every piece of team news, and the bookmaker's final risk adjustment. If you consistently take prices better than that, you were early on information the market later agreed with.
Compute it in probability terms, not odds terms, because odds are not linear:
CLV = (implied probability of your price) − (implied probability of the closing price). Positive means you beat the close. Implied probability = 1 / decimal odds.
Example: you take 2.10 (47.6%) and it closes at 1.95 (51.3%). Your CLV is −3.7 percentage points — you took a worse price than the market ended at, even if the bet won.
| Your price | Closing price | CLV | Verdict |
|---|---|---|---|
| 2.10 | 1.95 | −3.7pp | Took the worse side of the move |
| 2.10 | 2.10 | 0.0pp | No edge demonstrated |
| 2.10 | 2.30 | +4.1pp | Beat the close |
Building the record that makes CLV possible
CLV only works if you log the entry price, the closing price and the timestamp — every time.
Most people cannot compute their own CLV because they never recorded the closing price. Fix that first. The minimum viable log per bet is: fixture, market, line, entry price, entry time, closing price, closing time, stake, result.
Then evaluate in blocks. A hundred bets is the smallest sample worth reading, and you should look at the trend of CLV rather than a single block — an improving trend across blocks is stronger evidence than one good month.
The uncomfortable implication: if your CLV is consistently negative, the model is not working, regardless of your profit and loss. Positive results with negative CLV are luck that will revert.
Frequently asked questions
What is a good closing line value?
Consistently beating the close by 2 percentage points or more on a hundred-plus bets is a strong signal that your estimates carry information the market lacks. A single positive month means nothing on its own.
Can you have negative CLV and still profit?
Yes, over short samples. Random variance can produce profit while you are systematically taking bad prices, and that profit reverts. This is why professional evaluation weights CLV far above short-term results.
Does closing line value work for Asian handicaps?
Yes, and arguably better than for 1X2, because the Asian market carries lower margin and the closing line is a cleaner read on informed opinion. Record the line and price together, since the same price on a different line is a different bet.
See the model output for tonight's fixtures
Every match page shows the inputs, the reasoning and the confidence level — so you can audit the call instead of trusting a tip.
Open today's analysis