Closing line value explained for betting

In one sentence

Closing line value (CLV) measures how much better the odds you took were than the final market price just before the event started.

How it works

The closing price is the market's last and best-informed view. It has absorbed team news, weather, late money and the opinions of the sharpest bettors. If you regularly get bigger prices than the close, you are usually getting value, because you are ahead of that final consensus.

Results take thousands of bets to separate skill from luck. CLV is far less noisy, because every bet gives you a price comparison whether it wins or loses. A few hundred bets with consistent positive CLV is stronger evidence than a few hundred bets of profit.

On Betfair, the last traded price or the Betfair Starting Price (BSP) makes a good closing benchmark, since exchange books are close to 100% already.

The maths

\text{CLV} = \frac{O_{\text{taken}}}{O_{\text{close}}} - 1

EV \approx p_{\text{close}} \times O_{\text{taken}} - 1, \quad p_{\text{close}} = \frac{1}{O_{\text{close, fair}}}

O taken: the odds you actually got.
O close: the closing odds (BSP or final exchange price).
O close, fair: the closing odds with any margin removed.
p close: the closing fair probability, treated as the best estimate of the true chance.

In plain English: if the closing price is right, your edge is roughly how much bigger your price was than the close.

Worked betting example

Football, Over/Under goals. You back Over 2.5 goals in Leeds v Burnley on Betfair at 2.40 in the morning. It closes at 2.10 at kick-off. Commission is 2%.

CLV = 2.40 ÷ 2.10 − 1 = 0.143, so +14.3%.
Treat the close as fair: p close = 1 ÷ 2.10 = 47.6%.
EV before commission = 0.4762 × 2.40 − 1 = +14.3%.
Effective odds after commission = 1 + 1.40 × 0.98 = 2.372.
EV after commission = 0.4762 × 2.372 − 1 = +12.95%.

A single bet proves nothing. But if your average CLV across 500 bets is steadily positive, it is strong evidence that your process finds prices the market later agrees were too big.

If you bet at bookmakers, remove the margin from closing prices first (see Margin removal), or you will understate your CLV.

Where it's good

Judging a model or tipster long before results become meaningful.
Spotting when an edge is fading: CLV often drops before profits do.
Comparing when to bet: morning versus late, for instance.
Pre-match football, where the closing price is well formed.
Setting targets for automated strategies: a minimum CLV to keep a rule live.

Limitations and pitfalls

It assumes the closing price is the best estimate of truth. That is broadly true in the major leagues; check it holds in any league you rely on.
CLV and actual EV are not the same thing. Positive CLV with negative profit over a long run means something else is wrong, such as commission or settlement.
In-play and trading strategies have no simple closing line.
Bookmaker closing prices carry margin; comparing raw prices gives misleading CLV.

How to build it

pandas to join your bet log to closing prices by market and selection ID; Betfair historical data or BSP files for closes.
Data: exact odds taken, time of bet, and a consistent closing benchmark.
Practical tip: track CLV as the ratio of your odds to the close, and also as the difference in implied probability. The second is less distorted by longshots.

Market efficiency: why the close is a good benchmark.
Margin removal: fair closing prices from bookmakers.
Expected value: CLV is an estimate of EV.
Model vs market: testing your model against closing prices.
Hypothesis testing: judging whether average CLV is above zero.