CLV Calculator

Work out your closing line value, and whether the sample you have is large enough to mean anything yet.

This measures how your prices compared with the market's closing prices and how much confidence that comparison supports; it does not measure your profit, and a positive figure here is evidence about your bet selection rather than a forecast of your results.

Average closing line value, margin removed

+0.50%

95% interval −1.39% to +2.40%

The interval includes zero, so a bettor with no edge could produce this sample without anything unusual happening.

Short of the closeFair closeBeat the close

−10%0%+10%

Average95% intervalOne dot is one bet

Bets read: 16 bets
Middle bet: +0.72%
Spread, one deviation: 4.02%
Beat the close: 9 of 16
Landed on the close: 0
Measured against: Fair close

Your bets, one per lineEach line is the price you took, then the price that selection closed at. Add a third price, the close on the other side of the same market, and the margin is removed before your value is measured. American prices look like -110 or +105. Commas, tabs or spaces all work, so a column pasted from a spreadsheet reads as it is.

Everything is computed in your browser. Nothing you enter is stored or sent anywhere.

The same bets, measured the other way

Against the raw close

+4.38%

95% interval +2.41% to +6.33%

Against the fair close

+0.50%

95% interval −1.39% to +2.40%

Against the raw close these bets read +4.38%. Once the margin comes out, the same bets read +0.50%, about 11% of the raw figure surviving. The book's margin alone cut the apparent edge by 3.88 percentage points.

Beat rate moves too, and not by the same amount: 81.3% of these bets beat the raw close, 56.3% beat the fair one. The two statistics have different variance behavior, so a change in one does not size the change in the other.

Beat rate, a different statistic

Beat the close

56.3%

95% interval 33.2% to 76.9%

This is the share of bets that beat the close at all, whatever the size of the gap. It is a different quantity from the average above, with different variance behavior, so the two never share an interval: a bettor can beat the close on most bets by tiny amounts and still average negative, or beat it on fewer than half and average strongly positive.

How much more evidence it would take

12 to 270

more bets, at the spread these ones show

Resamples at or below zero

32%

That is the middle half of the projection taken across resamples of your own bets, not a single figure, because the average it is built on is itself uncertain. In 32% of those resamples the average came out at or below zero.

Read it as a statement about how much evidence a sample of that size would carry, not as a target. Betting more in order to find out whether you are ahead is how a small suspected edge gets funded by a real loss, and the projection assumes the average and the spread you have now both hold, which is the assumption a bettor with a short record is least entitled to.

Bet by bet

Taken	Closed	Fair close	−10%0%+10%	Value
+106	-110	+100		+3.00%
+151	+150	+157		−2.49%
-115	-140	-128		+5.08%
+228	+220	+231		−0.94%
+105	-105	+104		+0.28%
+101	+105	+114		−6.03%
-168	-200	-180		+2.55%
+323	+300	+317		+1.52%
-116	-120	-109		−2.85%
+210	+180	+190		+7.02%
-166	-160	-148		−4.45%
+147	+125	+133		+5.93%
-102	-115	-104		+1.17%
+252	+260	+272		−5.43%
-147	-175	-162		+3.96%
+118	+110	+119		−0.32%

Closing line value is evidence about the prices you took, not a record of what you won. Beating the close over a stretch you lost money on is an ordinary result rather than a contradiction, and the reverse is equally ordinary. Read why the closing line is used as the scoreboard or price a single bet before you take it.

Written and reviewed by Alexander Korsager · last reviewed October 6, 2026 · corrections

The idea

Closing line value is the gap between the price you took and the price the market settled on. If you took +110 on a selection that closed at -110, you got a better price than the market's final estimate, and that gap is what CLV measures. It is used as a proxy for expected value because it arrives long before results do: a bettor's profit and loss is mostly noise for hundreds of bets, while the closing line is the sharpest estimate the market produces and is available on every bet, won or lost.

The arithmetic is one division, which is why every free CLV tool returns a single percentage and stops.That single percentage is the problem. Per-bet CLV is spread out and skewed, so an average taken over a few dozen bets can land comfortably above zero for someone with no edge at all. Without knowing how wide the spread was, a bettor has no way to tell those two situations apart, and the usual rules of thumb ("+1% to +2% is strong", "60% beat rate over 200 bets") do not distinguish them either.
There is a second problem that the single percentage hides.A closing price on its own still contains the book's margin, so a raw CLV figure is measured against a number that is not the market's estimate of the true chance. Beating a vigged close by a little can be break-even or worse. Enter the closing price on the other side of the same market and this page removes the margin before measuring, and shows both readings for your own bets side by side, so you see exactly how much of your apparent edge the margin alone was worth rather than taking a rule of thumb's word for it.
So this page gives you three things instead of one: your average CLV with a confidence interval around it, your beat rate with its own separate interval, and a projection of how much more evidence it would take before the average could be distinguished from zero. Often the honest answer is that you cannot tell yet from this many bets, and the tool says so rather than rounding it into encouragement. Everything is computed in your browser. Nothing you type is stored or sent anywhere.

FAQ

Questions players ask

How much closing line value actually means something?

Less than the round numbers in circulation suggest, and it depends entirely on your sample size and how spread out your per-bet CLV is. An average of +1% over 40 bets whose values swing between -8% and +9% is well inside what no edge at all produces; the same +1% over 2,000 bets is not. That is why this page prints an interval instead of a figure. The interval is the answer to this question for your bets specifically, rather than a rule of thumb that ignores the two things that decide it.

How many bets do I need before CLV tells me anything?

There is no fixed number, and the widely quoted figures of 100 or 200 bets are not derived from anything. The sample size you need falls out of the size of your average relative to its spread: a large edge with tight dispersion is visible quickly, a small edge with wide dispersion may need thousands of bets. This tool projects a range from the dispersion your own bets show. Treat that projection as a statement about how much evidence a sample of that size would carry, not as a target to go and reach.

Should I measure against the raw closing price or a de-vigged one?

A de-vigged one, and enter the opposing price on every line rather than most of them. The tool removes the margin only when all of the bets carry a second closing price, so one gap puts the whole sample back on raw closes. The raw closing price includes the bookmaker's margin, so measuring against it flatters every bet. Two sides closing at -110 imply 52.4% each, which sums to 104.8%; removing that margin proportionally puts the fair price at even money. Once every line carries an opposing price, this page prints your own bets both ways, side by side, so you see the actual size of the difference on your own sample rather than a single made-up bet.

What is the difference between average CLV and beat rate?

Average CLV is the mean size of the gap; beat rate is the share of bets where the gap was positive, regardless of size. They are different statistics with different variance behavior, and a confidence statement about one says nothing about the other. A bettor can beat the close 70% of the time by tiny amounts and lose to it heavily on the rest, or beat it on only 45% of bets and still average strongly positive. This page computes an interval for each one separately and never presents either interval as covering both.

Does this tool store my betting history?

No. There is no account, no saved list and no upload. Everything runs in your browser, the numbers exist only while the page is open, and nothing is transmitted anywhere. The statistic needs only the price you took and the price it closed at, so the tool does not ask for your stake, your result, the date or the sport, and cannot record what it never receives.

Methodology

How this is calculated

Per-bet CLV is the price taken divided by the closing price, minus one. When every bet carries a second closing price, the market's margin is removed proportionally first (fair price = 1 + close / opposing close) and CLV is measured against that fair price instead.

That switch is all or nothing: if a single line is missing its opposing price, every bet is measured against the raw close, because a mean blending de-vigged and vigged figures averages two different quantities and would move when you added one price to one row.

The page names which of the two readings it used, on the headline label and again in the summary row beneath it, and the per-bet table gains a fair-close column only when the margin actually came out.

The headline is the mean of the per-bet figures, with a 95% interval from 2,000 seeded bootstrap resamples: each resample draws the same number of bets with replacement and the interval is the 2.5th and 97.5th percentiles of the resampled means.

Beat rate, the share of bets strictly beating the close, gets a Wilson score interval instead, which is the standard interval for a proportion.

The bets-to-significance projection solves n = (1.96 x standard deviation / mean) squared, run across the same resamples so its range reflects how uncertain the mean itself is. That range is the middle half of those resamples, the 25th to 75th percentile, rather than a 95% interval like the two above it.

Common misreadings4

An interval that clears zero is evidence about the prices you took, not a measurement of profit. Beating the close consistently and still losing money over the same stretch is an ordinary outcome, not a contradiction.
The bets-to-significance projection is not a plan. It says how much evidence a sample of that size would carry, and placing more bets to find out whether you are winning is how a small suspected edge gets funded by a real loss.
A positive raw CLV figure is measured against a price that still contains the margin, and one line missing its opposing price puts every bet on that reading, not just the line that is short. Small positive numbers there are the normal result of betting into a vigged market, not a finding. The worked example is the size of it: the same sixteen bets average +4.4% against the raw closes, with an interval that clears zero, and +0.5% once the margin comes out, with an interval that does not.
The interval assumes the bets entered are the whole set. Entering the ones you remember, or the stretch you are curious about, produces a confident-looking interval around a number that was chosen rather than observed.

Assumptions4

The closing prices you enter are from a market that actually closed and were the last prices available, not prices taken from part-way through the move.
Your bets are independent draws from one stable process. Bets on correlated selections, or a stretch of bets in one sport during one period, carry less information than the same count of unrelated bets.
The bets entered are all of them over the period, not the ones you remembered. A sample chosen after the fact is a different sample.
Proportional de-vigging splits the margin evenly across the implied probabilities. Other methods (power, Shin) shorten favorites and longshots differently, and the no-vig calculator shows how much that choice moves the fair price.

Limits7

A bootstrap resamples the bets it was given, so it can describe the uncertainty inside your sample but cannot correct a sample that is not representative of your betting.
Below 10 bets no interval is printed at all, because with that few values a percentile interval is mostly an artefact of which one or two bets happened to be extreme.
A sample where every bet's CLV comes out identical also prints no interval, however many bets it has. A bootstrap only ever resamples the values it was given, so resampling a dozen copies of one number returns that same number every time, and the zero-width interval that would produce reads as certainty rather than as the artefact it is.
The projection uses the normal standard-error identity even though the interval itself is a bootstrap. That approximation is small next to the uncertainty in the mean, which is why the projection is given as a range rather than a single number.
Past a projected 100,000 more bets, the page stops printing a number and says so instead. At that distance the honest statement is that the edge is too small next to its own noise to confirm from any sample a person would actually place, not a precise count to work toward.
Closing lines differ between books. Measuring against a soft book's close where you bet a sharp one, or the reverse, changes the answer, and the tool has no way to know which close you entered.
The stake on each bet is not an input, so this is an unweighted average across bets. A bettor whose stakes vary widely has a different money-weighted figure.

Evidence

The de-vig, the bootstrap, the Wilson interval and the sample-size identity are each derived in math.ts.

The fair price is asserted against the no-vig calculator's own proportional method, the Wilson interval and the sample-size identity are pinned against their closed forms, and the sixteen-bet worked example is pinned in both readings in math.test.ts: it clears zero measured against the raw closes and does not once the margin is removed.

A passing test means the code matches the model, not that your sample of bets is representative of your betting.

Gambling can cause serious harm. The National Problem Gambling Helpline is free, confidential and open 24/7, call or text 1-800-MY-RESET (1-800-697-3738), or chat at ncpgambling.org/chat.

Read up

Go deeper

Related tools