What is the vig? How sportsbook margin works, what books actually keep, and how it compares with an exchange fee
Where the vig comes from
A sportsbook is a market maker that quotes both sides of every event. If it quoted fair odds, +100 on each side of a coin flip, it would break even over time and go out of business on its costs. So it shades both prices. On the standard American line a $100 win costs $110 in risk, which converts to a decimal price of 1.909 and an implied probability of 52.38 percent. Two outcomes that must between them happen cannot each be 52.38 percent likely; the extra is not a forecast, it is the fee. Books call it the vig or juice, the industry's accountants call it hold, and European bookmakers describe the same thing as the overround, the amount by which the implied probabilities of all outcomes exceed 100 percent.
The two measures are related but not identical, and the difference trips people up. The overround at -110 each way is 4.76 percent, because 104.76 minus 100 is 4.76. The hold, the share of money wagered that the book keeps on a balanced book, is the overround divided by the total: 4.76 divided by 104.76, or 4.55 percent. Both are correct descriptions of the same margin; one is relative to fair probability, the other to handle. When a regulator publishes a "win percentage" or "hold," it is the second kind, measured on actual results rather than on the price.
The margin is not fixed at -110. Some books price major markets at -105 on each side, a 2.38 percent hold, to attract volume; most price player props, same-game parlays and futures far wider. A line of -120 on one side and +100 on the other holds 4.35 percent. A futures market in which the implied probabilities of every team add to 130 percent holds about 23 percent. The number that matters is always the sum of the implied probabilities across all outcomes, which is why the first step in reading any line is to convert each price and add them up; our implied probability guide covers the conversions and the odds converter does the arithmetic.
A worked example: $110 on each side
Take a point spread priced at -110 on both teams. Two bettors each risk $110, one on each side. The book now holds $220. One of them wins and is paid $210, the $110 stake back plus $100 profit; the other loses $110. The book's profit is $10 on $220 of handle, 4.55 percent, and it did not matter which team covered. That is the whole business model: balance the money, or at least balance the risk, and the vig does the rest.
Now look at it from the bettor's side. To break even over many bets at -110, a bettor must win 52.38 percent of them, since that is the implied probability of the price. A bettor who picks winners at exactly 50 percent, which is what a fair spread is designed to produce, loses the vig: on 100 bets of $110, 50 wins pay $5,000 in profit and 50 losses cost $5,500, a net loss of $500 on $11,000 wagered, 4.55 percent again. The margin is small on any one bet and relentless across many, which is why the hold figures regulators publish are so stable from year to year.
| Price on each side | Implied probability per side | Total (overround) | Hold on balanced book | Break-even win rate |
|---|---|---|---|---|
| +100 / +100 | 50.00% | 100.00% | 0.00% | 50.00% |
| -105 / -105 | 51.22% | 102.44% | 2.38% | 51.22% |
| -110 / -110 | 52.38% | 104.76% | 4.55% | 52.38% |
| -120 / +100 | 54.55% / 50.00% | 104.55% | 4.35% | 54.55% on the favourite |
| -115 / -115 | 53.49% | 106.98% | 6.52% | 53.49% |
What sportsbooks actually keep
The vig on a single spread bet is the floor, not the average. Regulated books report their win as a share of handle to state regulators, and the figures run well above 4.55 percent for three reasons: they sell products with wider margins, bettors do not split their money evenly, and favourites and overs attract more public money than their prices justify, so an unbalanced book sometimes wins big. Nevada's figures, compiled by the UNLV Center for Gaming Research from Gaming Control Board reports, show the state's sportsbooks won $601.4 million on $8.07 billion wagered in calendar 2025, a hold of 7.45 percent. Football held 7.98 percent, basketball 5.88 percent, baseball 7.18 percent, hockey 8.50 percent and parlay cards 32.59 percent. New Jersey's Division of Gaming Enforcement figures for 2025, as reported by Covers, put the state's operators at $1.18 billion of revenue on just over $12 billion of handle, a 9.7 percent hold, up from 9.6 percent in 2024, with December 2025 alone at 11 percent.
The gap between Nevada and New Jersey is itself a lesson. Nevada's market is older, more retail and more weighted to straight bets; New Jersey's is almost entirely online and dominated by operators that push parlays and same-game parlays, which carry the highest margins on the menu. A book's published hold is therefore a statement about its product mix as much as about its pricing, and a bettor who sticks to straight bets at -110 is paying something closer to the floor than to the state average.
| Market, 2025 | Handle | Win | Hold |
|---|---|---|---|
| Nevada, all sports | $8.07 billion | $601.4 million | 7.45% |
| Nevada, football | $2.55 billion | $203.5 million | 7.98% |
| Nevada, basketball | $2.46 billion | $144.9 million | 5.88% |
| Nevada, parlay cards | $18.8 million | $6.1 million | 32.59% |
| New Jersey, all sports | Just over $12 billion | $1.18 billion | 9.7% |
Why parlays multiply the vig
A parlay pays the product of its legs' decimal odds, and so it also compounds their margins. Two legs at -110 pay 1.909 times 1.909, or 3.64 to 1 including stake, where two fair coin flips would pay 4.00; the book's hold on that ticket is 8.9 percent, nearly double a single bet. Three legs pay 6.96 against a fair 8.00, a 13.0 percent hold. Four legs pay 13.28 against 16.00, a 17.0 percent hold. The bettor who finds -110 acceptable on one game is accepting a four-game parlay priced as if the vig were almost four times higher, which is why parlays are the product books promote hardest and why Nevada's parlay cards hold more than four times what its football straight bets do. Our parlay calculator shows the compounding leg by leg.
How an exchange fee compares
A prediction market prices the same game differently. A contract on the favourite that trades at 58 cents for yes implies 42 cents for no, and the two sides add to $1: there is no overround, because the exchange does not set the price, the order book does. The venue earns a fee on the trade instead. Under Kalshi's published schedule, effective July 7, 2026, a taker pays 0.07 times the number of contracts times the price times one minus the price, rounded up, which is 1.75 cents on a 50-cent contract, 1.68 cents at 60 cents and 0.63 cents at 90 cents; a maker whose resting order is filled pays nothing on most series, and some series carry a zero taker multiplier. Polymarket US charges sports takers 5 percent of the same price-times-one-minus-price term, a maximum of 1.25 cents per contract, as described in our fee comparison.
Put on the same footing as the sportsbook, a 50-cent Kalshi contract bought as a taker costs 1.75 cents on a 50-cent stake, 3.5 percent of the stake, against the 10 percent premium a -110 bettor pays over fair; expressed as a hold on the round trip it is about half of the sportsbook's 4.55 percent, and zero for a maker. That comparison holds for a two-way market with a tight order book. It does not hold automatically, because the exchange has a cost the fee formula does not show: the spread between the best bid and the best ask, which can be several cents on a thin market and is paid by anyone who needs to trade immediately. On a liquid NFL game the spread is often one cent; on a mid-week college game it can be wider than the sportsbook's vig. The honest comparison is fee plus spread against vig, market by market, which is what our price comparison tool is built to do.
| Venue | How the margin is charged | Cost on a 50/50 market | What it does not include |
|---|---|---|---|
| Sportsbook at -110 / -110 | In the price; both sides overpriced | 4.55% of handle on a balanced book | Wider margins on props, parlays and futures |
| Kalshi, taker | Fee per contract: 0.07 × price × (1 − price) | 1.75 cents per $1 contract, 3.5% of a 50-cent stake | Bid-ask spread; non-standard multipliers on some series |
| Kalshi, maker | Maker multiplier 0 on most series | 0 | Spread; the risk of not being filled |
| Polymarket US, sports taker | 5% × price × (1 − price) | 1.25 cents per $1 contract | Spread; category caps differ |
Reading the vig in practice
Three habits follow from the arithmetic. First, add up the implied probabilities before comparing two books; a favourite at -150 means something different next to +130 than next to +120, and the second book is charging almost twice the margin. Second, treat the published hold as a product-mix number: a state's 9.7 percent does not mean a straight bettor paid 9.7 percent, and a book's 4.55 percent spread price does not mean its customers paid only that. Third, when weighing an exchange against a sportsbook, count the fee and the spread together and remember that a maker on an exchange pays no fee but is not guaranteed a fill. The vig is not hidden. It is published in every line, and the only work is to read it.
What this guide does not cover
This guide explains the margin on two-way and multi-way markets and the simplest hold calculation. It does not cover the de-vigging methods that assign more of the margin to longshots, the pricing of same-game parlays with correlated legs, or how books move lines to balance risk rather than money. Hold figures are calendar-2025 totals as published by the UNLV Center for Gaming Research from Nevada Gaming Control Board reports and as reported by Covers from New Jersey Division of Gaming Enforcement data; monthly figures swing with results and later years will differ. Exchange fees are from Kalshi's schedule effective July 7, 2026 and the Polymarket US sports rate in our fee comparison as of October 2026; both change. The sportsbook prices are illustrative. Nothing here is legal, tax, financial or betting advice, and nothing here recommends placing any bet or trade.
Frequently asked questions
What is the vig in sports betting?
The vig, short for vigorish and also called juice, margin or hold, is the amount a sportsbook builds into its odds so that it pays winners less than fair value. On a standard point spread priced at -110 on each side, a bettor risks $110 to win $100; if the book takes $110 on both teams it collects $220 and pays out $210, keeping $10, or 4.55 percent of the money wagered. In implied-probability terms the two sides add up to 104.76 percent, and the 4.76 points above 100 are the overround.
How much do sportsbooks actually keep?
More than the vig on a single spread bet, because books also sell parlays, props and futures with higher margins and because results do not always balance. Nevada's regulated sportsbooks kept 7.45 percent of the $8.07 billion wagered in calendar 2025 according to the Gaming Control Board's annual figures: 7.98 percent on football, 5.88 percent on basketball and 32.59 percent on parlay cards. New Jersey's operators held 9.7 percent of just over $12 billion in 2025, up from 9.6 percent in 2024.
Is the vig the same as the overround?
They measure the same margin two ways. The overround is the amount by which the implied probabilities of all outcomes exceed 100 percent; at -110 each way it is 4.76 percent. The hold, or vig as a share of handle, is the margin divided by the total implied probability; at -110 each way it is 4.76 divided by 104.76, or 4.55 percent. Both describe what a book keeps if it attracts balanced money.
Do prediction markets have a vig?
Not in the price. On an exchange such as Kalshi or Polymarket US the yes and no prices of a contract add up to $1, so there is no overround. Instead the venue charges a fee on the trade: Kalshi's taker fee is 0.07 times price times one minus price per contract, a maximum of 1.75 cents at a 50-cent price, and makers pay nothing on most series. The equivalent cost is usually lower than a sportsbook's vig on a two-way market, but the bid-ask spread and thin order books can add a cost the fee formula does not show.
Related
- Odds converter: American, decimal, fractional and implied probability
- Parlay calculator: how the edge compounds across legs
- Implied probability explained: odds to percent, vig out
- Sportsbook vs prediction market: how the two models price the same game
- Is Kalshi legal in your state? The 50-state tracker
Sources
- UNLV Center for Gaming Research, Nevada Gaming Win 2025, compiled from Nevada Gaming Control Board monthly revenue reports, for the statewide sports book handle, win and win percentage by sport.
- Covers, New Jersey sportsbooks generate $1B in December handle, $12B in 2025, 19 January 2026, reporting New Jersey Division of Gaming Enforcement figures for annual handle, revenue and hold.
- Kalshi, Fee schedule, effective 7 July 2026, for the taker and maker formulas, the rounding rule and the non-standard multipliers.
- Kalshi Help Center, How are prices calculated?, for yes and no prices summing to $1.
- BetG8, Kalshi vs Polymarket fees compared, for the Polymarket US sports rate.
18+ only. Betting and prediction market contracts carry real financial risk and you can lose what you stake. This page explains arithmetic; it is not legal, tax, financial or betting advice. If gambling stops being fun, help is available at 1-800-GAMBLER.