Good CLV is a positive average closing line value, measured against the no-vig closing price, held over enough bets that it can't be put down to noise. As a rough guide, an average of +1% to +3% over a few hundred bets is a solid sign that you're taking better prices than the market settles on. But there's no published standard for "good", so the better test is your own number against its own standard error. Below zero against a fair close is bad CLV. So is zero against a close that still has the bookmaker's margin in it.
This post assumes you know what CLV is. If you don't, start with our closing line value explainer, then come back for the benchmarks.
Why there's no single "good CLV" number
You'll see bands quoted everywhere: +1% is decent, +3% is sharp, +5% is elite. Those are rules of thumb. The same CLV number can mean very different things depending on three choices you made when you measured it:
- Which close. One bookmaker's last price, the best price across books, a median, or a fair price with the margin removed.
- Which formula. Odds ratio or difference in implied probability. They give different numbers for the same bet.
- Which bets. NBA head to head closes are tighter than NBA player props, and both behave differently from racing.
So before you judge your CLV, fix how you measure it. Everything below uses one formula, the same one as our explainer:
CLV % = (your odds ÷ fair closing odds) - 1
Measure against the no-vig close
Here's why the close you pick matters so much. Invented numbers, an NBA head to head:
- You took Team A at $2.10.
- At tip-off both teams are $1.95.
- Each $1.95 implies 51.28%. Together that's 102.56%, so the market's margin is 2.56%.
- Take the margin out and each side is 50%, so the fair closing odds are $2.00. Our no-vig calculator does this step for any two-way or multi-way market.
Now the CLV:
| Measured against | Sum | CLV |
|---|---|---|
| Raw close ($1.95) | 2.10 ÷ 1.95 - 1 | +7.7% |
| No-vig close ($2.00) | 2.10 ÷ 2.00 - 1 | +5.0% |
The raw close flatters you by about 2.7 points, purely because the margin is still in it.
It works the other way too. Suppose you'd taken $1.95 and the line never moved. Against the raw close that's 0% CLV, which sounds like break-even. Against the fair close it's 1.95 ÷ 2.00 - 1 = -2.5%. That's the margin you paid.
There's a useful property hiding here. If the fair close is an accurate estimate of the true chance, then CLV against it is roughly your expected return per bet. So +5.0% against a fair close is about +5% expected, while +5.0% against a raw close might be close to nothing.
What real data shows
Economist Karl Whelan tested this on NBA moneylines over three seasons: 3,670 games, quotes from about ten licensed US books, with the closing line taken as the median of the final quotes. About half of all bets had positive CLV, and bets with positive CLV did better than bets with negative CLV. But only the top tenth by CLV made a meaningful profit, and the ninth tenth averaged 5% CLV for a 0.2% return. The average loss rate across his bets was about 4.5%, roughly the market's margin.
That ninth-decile result looks odd until you put the margin back in. A close with a margin near 4.5% in it overstates CLV by about that much, and 1.05 ÷ 1.045 - 1 leaves roughly half a per cent against fair odds, which is close to what those bets returned. Whelan's own conclusion is blunter: positive CLV alone isn't proof of profit, and you need to beat the margin, not just the closing line.
Racing: use BSP as the close
For racing, the cleanest close is the Betfair Starting Price. BSP is set by a reconciliation algorithm that balances back and lay orders as the market suspends, and Betfair's own data scientists describe it as a very strong estimate of a runner's chance, though not perfect. It has no bookmaker margin built in. Compare your fixed-odds price with BSP and keep that rule for every race.
Good, bad and too good: reading your average
Measured against a no-vig close, after enough bets to trust it (next section), this is a reasonable way to read the number. The bands are guidance, not a published standard.
| Average CLV vs fair close | How to read it |
|---|---|
| Below 0% | Bad CLV. You're taking worse than the fair closing price on average. Common causes: betting after news has moved the line, or betting at the books with the biggest margins. |
| 0% to +1% | Hard to tell apart from zero unless the sample is very large. |
| +1% to +3% | A solid sign your prices beat the market's final view, if it holds over a few hundred bets. |
| +3% to +6% | Strong. Worth checking your data before trusting it. |
| Above +6%, sustained | Check for errors first: stale closing prices, closes taken from a slow book, prices logged that you didn't actually get, or bets with missing closes quietly dropped. |
Two things to check when the number looks great:
- Missing closes. Decide up front what happens to a bet with no closing price and apply it to every bet. BetBank's public track record leaves those picks out rather than counting them as zero, and says so on the page. It also notes that the no-skill baseline for its CLV figure isn't published yet, which is the honest way to show the number.
- Correlated bets. Five player props from one NBA game aren't five independent bets. If one injury report moves them all, they count for something closer to one.
Beat-close rate is a different number
Your beat-close rate is the share of bets where you got a better price than the close. Your average CLV is how much better or worse on average. They can point in different directions.
Ten invented bets: four beat the fair close by 8%, six miss it by 1%.
- Beat-close rate: 4 of 10 = 40%
- Average CLV: (4 × 8 - 6 × 1) ÷ 10 = +2.6%
A beat-close rate under 50% with a positive average is common for bettors who wait for big mispricings. Track both, but judge your prices on the average.
How many bets before your CLV means anything
This is where CLV beats profit as a yardstick. Each bet's CLV is a small number that doesn't swing much. Each bet's profit is a full win or a full loss.
The tool is the standard error: the standard deviation divided by the square root of the number of bets. Work out the standard deviation of your own per-bet CLV (one spreadsheet formula on your CLV column; our bet tracking spreadsheet guide shows the columns). Then:
Standard error = per-bet spread ÷ √(number of bets)
If your average is more than about two standard errors above zero, it's unlikely to be noise. Under a normal curve, about 95% of results fall within two standard deviations.
CLV settles much faster than ROI
Invented but realistic: per-bet CLV spread of 6 percentage points on NBA sides. For ROI at $2.00 odds, each bet returns +1 or -1 unit, and with a win chance near 50% the Bernoulli variance gives a per-bet spread of about 100 percentage points.
| Bets | Standard error of average CLV (6-pt spread) | Standard error of ROI at $2.00 |
|---|---|---|
| 25 | 1.2 pts | 20 pts |
| 100 | 0.6 pts | 10 pts |
| 400 | 0.3 pts | 5 pts |
| 1,000 | 0.19 pts | 3.2 pts |
After 400 bets your ROI could be 10 points either side of its true value. Your average CLV is pinned to within about 0.6 points. That's why CLV is the number to judge your prices on over a season, and why our guide to betting ROI and sample size warns against reading too much into a short run of results.
Bets needed for a two-standard-error read
Flip the formula round: bets needed ≈ (2 × spread ÷ average CLV)².
| Average CLV | NBA sides (6-pt spread) | Player props (10-pt spread) |
|---|---|---|
| +0.5% | 576 | 1,600 |
| +1% | 144 | 400 |
| +2% | 36 | 100 |
| +3% | 16 | 45 |
Props get a wider spread here because their lines move in bigger jumps and their closing prices are less reliable. Use your own spread, not these. And remember the correlated-bets point: if many bets share a game, your real count is smaller than the rows in your sheet.
A quick routine for checking your CLV
- Log the odds you actually got, the time, and the market.
- At the close, record both sides of the market (or the full field) from the same reference each time.
- Remove the margin to get the fair closing price.
- Calculate CLV per bet, then the average, the beat-close rate and the spread.
- Divide the spread by √(bets). Only trust the average once it's about two standard errors clear of zero.
- Split it by sport and market. Good CLV on NBA sides and bad CLV on props is a finding, not a wash.
Practise on the NBA board
The NBA season is a good place to practise this: two-way markets, a game most nights, and closes you can de-vig in seconds on the Odds Screen. If you haven't had a paid BetBank plan before, the NBA season special makes a weekly or monthly Silver or Gold plan half price for your first 2 months (the first 2 monthly or first 8 weekly payments), after the 7-day free trial, if you start by 30 November 2026.
BetBank is for adults 18+. Every bet can lose, good CLV doesn't make a bet a winner, and no method or tool removes the risk. Set a limit, take breaks, and gamble responsibly. If it stops being fun, our responsible gambling page lists free, confidential support.