How Pinnacle's Closing Line Is Calculated (and De-Vig)
The closing line is not a number you look up. It is what is left after Pinnacle's margin is removed, and the de-vig method you choose can decide whether a bet has an edge at all.
Most articles about Pinnacle's closing line treat it like a number you look up. It is not. Pinnacle publishes odds, and odds contain margin. The closing line bettors actually measure against is what is left after that margin is removed. The removal step is called de-vigging, and the method you pick changes the result.
This matters because closing line value (CLV) is the only fast way to know if a betting strategy works. If you measure it against a sloppy version of the closing line, you get a sloppy answer. Here is how the line is built, how the margin comes out, and where people get it wrong.
Why Pinnacle, and what "closing" means
Pinnacle runs low margins (often around 2 percent on main football markets) and takes large stakes from winning bettors instead of limiting them. Sharp money moves their price. By the time a match kicks off, their odds have absorbed most of what the market knows.
The closing line is the last price Pinnacle offers before kickoff. Not the price an hour before, not the price when you bet. The last one. For pre-match bettors it is the closest thing to a public "true probability" that exists.
But it still has margin in it. If you add up the implied probabilities of every outcome in a Pinnacle market, you get more than 100 percent. That excess is the vig. A fair line is one where the probabilities sum to exactly 100 percent.
De-vig, step by step
Take a hypothetical Pinnacle line on a football match: home 2,10, draw 3,40, away 3,60.
Convert to implied probability by dividing 1 by the odds: 47,62 percent, 29,41 percent and 27,78 percent. Sum: 104,81 percent. The 4,81 points above 100 are the margin (this is a wider margin than Pinnacle typically runs, chosen to make the arithmetic visible).
Now remove it. There are several ways, and this is where it gets interesting.
Multiplicative (proportional). Divide each implied probability by the total. Home becomes 47,62 / 104,81 = 45,43 percent. Fair odds: 2,20 home, 3,56 draw, 3,77 away. Simple, and the method most free calculators use. Its weakness: it spreads the margin evenly by proportion, which is not how bookmakers actually price.
Additive. Subtract an equal share of the margin from each outcome (4,81 / 3 = 1,60 points each). Home becomes 46,02 percent. This punishes longshots less than the multiplicative method, but it can produce negative probabilities on markets with many outcomes, so it is rarely used seriously.
Power. Raise each implied probability to the same exponent k, and solve for the k that makes them sum to 100 percent. Here k is about 1,046. Home becomes 46,02 percent, draw 27,80 percent, away 26,18 percent. Fair odds: 2,17 home, 3,60 draw, 3,82 away.
Shin. A model that assumes part of the margin exists to protect the bookmaker from insiders, and that this protection is loaded onto longshots. Mathematically heavier, results usually land close to the power method.
The power and Shin methods exist because of the favourite-longshot bias: bookmakers put more margin on outsiders than on favourites, in probability terms. A proportional de-vig ignores this and makes favourites look slightly worse value than they are, and longshots slightly better.
We use the power method. It is well documented, it handles the bias, and it is stable across two-way and three-way markets.
The same price, two different answers
Say a bookmaker in your country offers 2,22 on the home side, and Pinnacle is at 2,10 as above.
With multiplicative de-vig, fair probability is 45,43 percent. Your edge at 2,22 is 2,22 x 0,4543 - 1 = 0,85 percent. Barely anything.
With power de-vig, fair probability is 46,02 percent. Your edge is 2,22 x 0,4602 - 1 = 2,16 percent. A real, if thin, edge.
Same bet. Same Pinnacle line. The de-vig method decides whether it clears a 2 percent threshold or not. Anyone who tells you their edge to one decimal without saying how they de-vig is not telling you much.
Measuring CLV against the fair close
Now the match is about to start. Pinnacle has moved to 2,00 home, 3,45 draw, 3,80 away. Power de-vig gives a fair closing probability of 48,24 percent on the home side, or fair closing odds of 2,07.
You took 2,22. Your CLV is 2,22 x 0,4824 - 1 = 7,1 percent.
Two things to notice. First, if you measured against the raw closing price of 2,00 instead of the fair 2,07, you would report 11 percent CLV. That is an inflated number, and it is the one many tipsters quote. Second, the line moved your way. That movement is the actual evidence. The market agreed with you after you bet, and the bet would have been priced lower if you had waited.
You can lose that bet and it is still a good bet. You can win it and it is still a good bet. The result does not enter the calculation.
Where our numbers come from
Every pick we publish is stored with the price at the time, Pinnacle's closing price and the fair probability after power de-vig. CLV is averaged across every pick with a recorded close, settled or not. Since 4 September 2026 the value engine has published 144 picks. Of those with a recorded close, the average CLV is +4,0 percent (n=119). Return on the settled picks is +13,7 percent, with a profit of €912,50 on €6.650 staked (n=133).
The ROI will move a lot over the next hundred bets. The CLV will move less, and it is the number we watch.
What this does not mean
A positive CLV does not mean every bet is good, or that the next month will be profitable. It means that on average, the market moved towards our prices after we published them, which is the standard evidence that a strategy is finding value rather than getting lucky.
Pinnacle's closing line is also not perfect. It is a market price with margin, not an oracle. In small leagues with low limits it can be slow to adjust, and for some markets Pinnacle offers no price at all, in which case a fair reference has to come from elsewhere (we use a betting exchange for those). Any single CLV figure carries noise. With n=119, a spread of a couple of percentage points either way is normal. Judge the number by the sample behind it, not the headline.
And a fair line is only as good as the de-vig behind it. If you build your own tracking, pick one method, state it, and never switch mid-sample.
Free picks daily in Telegram: t.me/thebetlab_free How we measure CLV, and the live numbers: thebetlab.io/#clv
What The Bet Lab is
A members club that publishes pre-match value bets and measures every one against Pinnacle's closing line. Every result is public, wins and losses.
See how it works · Free picks daily on Telegram