How to remove the vig from odds

The number on the board is not a fair price. It is a payout with a fee inside it. That fee is the vig — also called juice, margin, or overround. Until you strip it, two-way implied probabilities add to more than 100%, and any comparison you make is against a loaded number.

This page owns one job: how to remove the vig from odds. Convert both sides of a two-way market to implied probability, divide each by the two-way total, and you have no-vig — fair — odds. Fair probability, then compare. Where you find a takeable price after that is a different article.

If you still need the conversion itself — a decimal turned into a percentage — start with how to read implied probability, then come back.

The posted number is loaded

A bookmaker does not post two sides that add to a coin flip. Each decimal converts to an implied probability: 1 ÷ decimal. Add both sides. The sum sits above 100%. The surplus is the bookmaker’s built-in fee.

That fee is vigorish. Keep Wikipedia’s definition: the charge for accepting the wager, not a forecast of who wins. Books also call it margin. Pinnacle’s margin calculator is a public example of the same idea from the bookmaker’s side — convert the posted odds, see how much of the price is their charge.

The posted decimal is still the number you get paid if you bet it. Nothing here changes the ticket. What it changes is the probability you attach to that ticket. Treat 1 ÷ 1.91 as “this side is 52.4%” and you have read the loaded figure, not the market’s no-vig view.

People compare those loaded percentages to each other, or to a model, and call the gap value. That is a category error. You cannot price a bet against a number that already includes the house take and call the remainder edge.

Fair odds are the two-way market with that take removed so the probabilities add to 100%. They are a reference. They are not “true probability” in any deep sense. They are the same prices, restated without the fee. The posted number stays loaded. The stripped number is what you compare against.

Juice, overround, and house edge are the same object in this context: the extra percentage above 100%. We stay on two-way markets — moneyline, spread, total. A three-way card is a different sum. Do not strip two sides and ignore a third.

How to strip the vig

We use one method on two-way markets: proportional, or multiplicative, vig removal. Convert each side to implied probability. Divide each implied by the two-way total. Other methods exist. We do not use them here.

Four steps. Same order every time.

1. Implied probability for a side = 1 ÷ that side’s decimal odds.
2. Two-way total = implied A + implied B. This is above 1.00 (100%). The excess is the margin in the market.
3. Fair probability = that side’s implied ÷ the two-way total.
4. Fair odds = 1 ÷ fair probability.

Do both sides. They will now add to 100%. Stay on the same two-way market. Do not mix a moneyline with a spread, or one book’s A with another book’s B.

Posted: 1.80 and 2.00. No teams. Two sides of one market.

Implied A = 1 ÷ 1.80 = 0.5556 = 55.56%.
Implied B = 1 ÷ 2.00 = 0.5000 = 50.00%.
Two-way total = 1.0556 = 105.56%.
Fair A = 0.5556 ÷ 1.0556 = 0.5263 = 52.63%.
Fair B = 0.5000 ÷ 1.0556 = 0.4737 = 47.37%.
Fair odds A = 1 ÷ 0.5263 = 1.90.
Fair odds B = 1 ÷ 0.4737 = 2.11.

The posted pair added to 105.56%. After the split, they add to 100%. Side A was paying 1.80; the no-vig number on that side is 1.90. Side B was paying 2.00; the no-vig number is 2.11.

Posted	Implied	Fair probability	Fair odds
1.80	55.56%	52.63%	1.90
2.00	50.00%	47.37%	2.11
Total	105.56%	100%	—

If you bet 1.80 on A at that same book, you are taking a shorter price than the fair 1.90. That is the vig on your ticket. The method did not say A is a bad side. It said 1.80 is a loaded price.

Use the same four steps every time. Change the method after a losing week and you are no longer measuring. The proportional split assumes the margin is spread in line with each side’s implied probability. That is the method. It is not a claim that the book built the line that way on purpose.

Fair odds vs the price you can actually bet

Fair odds are a benchmark. They are not a quote you can take at the book that posted the loaded pair.

That book already took the margin. Asking them to pay 1.90 when they posted 1.80 is asking them to give the fee back. They will not.

You strip the vig on a reference two-way — usually a sharper, lower-margin market on the same selection — and you compare that fair probability to a decimal you can actually get elsewhere. The fair number is the ruler. The takeable number is the bet.

If the book you can use is still paying more than the fair decimal, the price is longer than the no-vig reference. That gap is where value starts. If they are paying less, you are buying a worse number than the stripped reference. Skip it.

Two mistakes sit next to this.

First: treating fair odds as a price the market “should” offer you. It will not. Recreational books post wider numbers on purpose. The fair figure is for comparison, not for negotiation.

Second: stripping the vig on the same book you intend to bet, then calling the result +EV. You have only restated that book’s own margin. You have not found a misprice. You need a reference and a takeable quote, and they are usually not the same card.

A price you cannot take is not a bet. Fair 1.90 at a book with no account, or a limit already gone, is a number on a screen. Compare against books you can actually use. Hold those accounts before you start hunting gaps.

What a no-vig calculator does (and does not)

The arithmetic above is short. It is also easy to mistype. The no-vig calculator is that four-step split for a two-way market. You enter both posted decimals. It returns the fair probabilities and the fair odds.

That is the whole job. It supports this page. This page does not replace the tool, and the tool page does not need another tour of its fields.

It does not pick a winner. Fair 52.63% is not a tip. It is a restated price.

It does not scan other books. One pair in, one stripped pair out. If you want the same math applied across a live board, that is a scanner, not a calculator.

It does not compute expected value. EV needs a probability and the decimal you will actually bet. The calculator gives you the first number. The second number is a quote at a book you can use.

It does not size a stake, hold an account, or record a ticket. Those are later steps.

Use it when you have one two-way in front of you and you want the no-vig number without doing the division by hand. Then leave it. The output is an input to the next comparison, not a decision.

Then compare, then record

You now have a fair probability. Compare it to the price you can take.

Expected value is EV = (p × decimal) − 1, where p is that fair probability and decimal is the quote you will actually bet. The derivation lives on how EV is calculated. If the result is positive, the takeable number is longer than the no-vig reference. If it is negative, you are paying the tax.

One comparison is a line. A process is many of them. The full workflow — reference, strip, scan, size, book — is how to find positive EV bets. This page stops at the strip.

Doing that by hand across a board does not stay current. A value bet scanner is the same comparison at volume: strip a sharp two-way, flag a slower book still paying more than the fair price.

After you take a number, write it down at the price you got. Closing line value is the next check: whether that ticket still beat the last efficient price.

A week of results will not tell you if the strip-and-compare step is working. A large sample of unique tracked tickets will. That is why the EV Kings performance guarantee uses a floor of 500 unique tracked value bets over at least 30 days. If you still show a net loss after that volume, the first subscription cycle is refundable. It does not cover betting losses.

Start Summer Pass — 14 Days Free
14 days free, no credit card. Then €49/mo for 2 months → €99/mo. Annual €69/mo billed €828.
100% first-cycle refund if you place ≥500 unique tracked value bets over ≥30 days and still show a net loss. Does not cover betting losses.