Closing line value, usually shortened to CLV, is the single most common answer to the question "how do I know if I am actually any good at this, or just lucky?" It is also explained badly almost everywhere, usually by someone selling you something. So here is the plain version, followed by our own numbers, including the ones that make us look bad.
A bookmaker's price moves between the moment it opens and the moment the event starts. The final price, the one available just before kickoff, is called the closing line. It is the market's last word, after everyone has bet, after the team news is out, after the sharp money has landed.
Closing line value is the difference between the price you took and that closing price. If you backed a team at 2.10 and it closed at 1.90, you got a better price than the market's final verdict. That is positive CLV. If you took 1.90 and it closed at 2.10, the market moved away from you, and that is negative CLV.
Crucially, this has nothing to do with whether the bet won.
Backed at 2.00, closed at 1.80: (2.00 / 1.80 - 1) x 100 = +11.1%.
Backed at 1.80, closed at 2.00: (1.80 / 2.00 - 1) x 100 = -10.0%.
Two things people usually track: the percentage of bets that beat the close, and the median CLV across all of them. The first tells you how often you are ahead of the market. The second tells you by how much, without one freak result distorting the picture the way an average would.
A win rate over fifty bets tells you almost nothing. Variance in betting is brutal, and a losing approach can look excellent for months. CLV cuts through that, because it measures the decision rather than the outcome.
This is also why serious bookmakers restrict accounts that beat the close. They are not tracking your profit. They are tracking whether you are ahead of their own line, because that predicts your future profit better than your past profit does.
Most articles on this topic stop before showing anything real. Here is ours. The basis is every settled bet we have published, 310 of them, split by market rather than pooled, because different market types carry different margins and mixing them hides the story. Yield is the return on the total amount staked.
| Market | Settled bets | Closing price captured | Beat the close | Median CLV | Yield |
|---|---|---|---|---|---|
| Totals, variable line | 133 | 64% | 38% | 0.0% | +4.4% |
| Moneyline | 111 | 97% | 49% | 0.0% | +2.0% |
| Totals, over and under 2.5 | 42 | 90% | 50% | +0.2% | -25.0% |
| Match result, three way | 24 | 83% | 65% | +1.6% | -2.2% |
Read the first and last rows together, because they are the point of this article. Our best closing line value market, match result, where we beat the close almost two times in three, lost money. Our worst one, variable line totals, where we beat the close barely more than a third of the time, is the most profitable thing we run.
Over our sample, closing line value did not predict profit. It pointed the wrong way, in both directions at once.
The middle rows make it worse rather than better. Over and under 2.5 in football gave the most ordinary closing line value numbers on the page: half the bets beat the close, median a whisker above zero, nothing to see. It also lost a quarter of everything staked and is the one market we have switched off. The metric never flagged it. The results did, sixty days later.
We would rather argue against our own headline than have a reader do it for us.
For variable line totals we only record a closing price when the book closes on the same number we bet. If we took Under 7.5 and it closed at Under 8.0, those are two different bets, so we record nothing rather than compare two different things. The effect is visible in the coverage column: 64% for that market against 97% for moneyline, where the line cannot move because there isn't one.
So the 38% is not a statement about variable line totals. It is a statement about the subset of them where the market did not move the number, and the cases where the market disagreed with us most are precisely the ones most likely to be missing from it. A filtered sample is not a representative one, and we would rather say that out loud than let a scary looking number do work it has not earned.
No, and we are not going to stop tracking it. But it is worth being clear about what it is. Beating the close measures whether you got a better price than one specific book's final word. That is a proxy for having known something the market had not priced yet, which is in turn a proxy for future profit. Two proxies deep, at a soft book, on a line that can move underneath you, there is a lot of room for the measurement to drift away from the thing it is meant to measure.
The useful framing we have landed on: closing line value is a fast signal and results are a slow one. When a fast signal and a slow signal agree, you have learned something. When they disagree, as ours do, the honest position is that you do not know yet, and you keep both running rather than quietly adopting whichever one currently flatters you.
The first version of this article showed the same variable line totals row, called it a signal that something in the approach was wrong, and said the results had confirmed it. That last clause was not true. The results said the opposite then and say it more clearly now, at a yield of +4.4% across 133 settled bets.
We think we know how it happened, and it is boring rather than sinister. The story of a metric catching a problem before the results do is the story everyone in this category tells, because it is flattering. We had a row that fit the shape of that story and we wrote the story, without going back to check the yield column sitting in the same database. The correction came from rerunning our own numbers for a different reason entirely and noticing they contradicted a published claim.
That is also the argument for keeping a public record we do not edit after the fact. Every settled bet is on it, wins and losses, which is what made this correction possible and what makes it checkable by someone who does not take our word for any of it.