How bookmakers set odds: where the opening number comes from, how the margin is built in, and why the line moves

Guide · Updated November 3, 2026 · 18+ · Not legal, tax or financial advice · By BetG8 Editorial · Sports

How bookmakers set odds: opening line, margin and movement — The margin is built into the price before the first bet is taken.

Short version: A betting line starts as an estimate of how likely each outcome is, usually from power ratings and a model, and then has a margin layered on top so that the two sides add up to more than 100 percent. At the standard -110 price on both sides of a point spread, that margin is 4.55 percent of every dollar wagered. The opening number is deliberately posted with low limits, the first bets come from professionals, and the line moves toward the price that the sharpest money will accept rather than toward a 50-50 split of the handle. Nevada sportsbooks kept between 5.1 and 7.5 percent of everything bet on sports in each year from 2018 to 2025. This guide explains the mechanics with a worked example. It is not betting advice.

LIVE MARKETPro Football 2027 ChampionTrade on Polymarket →

Where an opening line comes from

Every sportsbook price begins as a probability. For a point spread the oddsmaker, or more often a model, assigns each team a power rating, adjusts for home advantage, injuries, rest and weather, and produces an expected margin of victory. For a moneyline or a futures market the same process produces a win probability. Pinnacle, the bookmaker most often cited as the industry’s reference price, describes the job in its educational material as pricing a fair market first and taking a margin second. Las Vegas operator Circa, which posts the first college football lines of each week during the season, puts it more bluntly: the opening number is a starting point, and “the most respected professional bettors will dictate what the betting line is, not the bookmaker” (Circa Sports).

Only a few books open a market. Circa says the NBA, NHL and MLB lines are still opened offshore in most cases, and that other books post numbers close to the existing market rather than risk being picked off by arbitrage bettors who find a two-point gap between venues. The book that opens accepts low limits, often a few hundred to a few thousand dollars, in exchange for information: the first bets tell it whether its number is wrong. By the time a recreational bettor sees a line on a Sunday morning, it has usually been tested by professional money for several days.

The margin: how -110 becomes 4.5 percent

A fair coin has two outcomes at 50 percent each, which in American odds is +100 on both sides. No bookmaker offers that. The standard point spread price is -110 on each side, meaning a bettor risks $110 to win $100. Convert each side to an implied probability and the two add up to 104.76 percent; the extra 4.76 percent is called the overround or the vig. Expressed as a share of all money wagered, the margin is 1 minus 1 divided by 1.0476, which is 4.55 percent. That is what the book expects to keep if it takes the same amount on both sides.

The size of the margin varies by market. Low-margin books price major football and basketball spreads at -105 or -107, which is 2.4 to 3.3 percent; many US books price player props and small-market events at -115 or -120 per side, which is 6.5 to 8.3 percent. Pinnacle’s own note on margins makes the point that matters for a bettor: the margin is the hurdle rate. At -110 you must win 52.38 percent of equal-sized bets to break even; at -105, 51.22 percent; at -120, 54.55 percent.

Price on each side	Implied probability per side	Two sides add to	Margin on handle	Break-even win rate
-105 / -105	51.22%	102.44%	2.38%	51.22%
-110 / -110	52.38%	104.76%	4.55%	52.38%
-115 / -115	53.49%	106.98%	6.52%	53.49%
-120 / -120	54.55%	109.09%	8.33%	54.55%

A worked example: pricing one game

Suppose a model rates the home team 3 points better than the visitor after home advantage. The oddsmaker posts Home -3 at -110 and Visitor +3 at -110. The fair probability of each side covering is 50 percent; the book quotes 52.38 percent for each. If $110,000 is bet on each side, the book collects $220,000, pays $210,000 to the winners ($110,000 stake returned plus $100,000 in winnings) and keeps $10,000, which is 4.55 percent of the handle.

Now the moneyline on the same game. The model says the home team wins 57 percent of the time. A fair moneyline would be -133 for the home side and +133 for the visitor. The book adds its margin by shading both prices toward itself: Home -150 (implied 60.0 percent) and Visitor +130 (implied 43.5 percent). The two now add to 103.5 percent, a margin of 3.4 percent on handle. Notice the margin is not applied evenly: the favourite’s price has been pushed 3 points of probability, the underdog’s only 0.5. Books do this because the public bets favourites more than it should, a bias documented below, and the extra cost is charged where the demand is.

Step	Home team	Visitor	Total
Model win probability	57.0%	43.0%	100.0%
Fair moneyline	-133	+133	—
Posted moneyline	-150	+130	—
Implied probability of posted price	60.0%	43.5%	103.5%
Shade added	+3.0 pts	+0.5 pts	3.4% margin

Our odds converter does the American-to-probability step for any price, and the implied probability converter removes the overround from a two- or three-way market to recover the book’s fair estimate.

Why the line moves, and why books do not balance

The textbook story is that a bookmaker wants equal money on both sides so that the losers pay the winners and the margin is locked in regardless of the result. The evidence says that is not how it works. Economist Steven Levitt’s 2004 study of 19,770 wagers by 285 bettors across 242 NFL games in the 2001-02 season (Economic Journal) found that in nearly half of games at least two-thirds of the bets fell on one side, and that the book made no attempt to move the line to fix it. Bettors put 60.6 percent of their bets on favourites, and 68.2 percent on visiting favourites. Over the 21 seasons from 1980 to 2001, favourites covered the spread only 48.2 percent of the time and visiting favourites 46.7 percent. The book, in other words, set lines that made the popular side slightly worse than a coin flip and let the imbalance stand. Levitt estimated that this raised the bookmaker’s gross margin from the 5.0 percent a balanced -110 book would earn to about 6.16 percent, a 23 percent increase.

What does move the line is money the book respects. Circa describes the process directly: when a bettor it knows to be sharp hits the opener, the book moves to a number that either splits the professionals or stops them, and large recreational bets are often ignored because they carry no information about where the spread should be. In the largest markets, such as the NFL, this happens almost in real time across the industry, which is why a line that opens at -3 on a Sunday evening can be -4.5 everywhere by Monday afternoon with no injury news at all. The closing line, the last price before kickoff, is the one that has absorbed the most information, which is why professionals measure themselves against it; our guide to closing line value covers that measurement.

What the hold data shows

The margin is a target; the hold is what the book actually keeps after results come in, and it is published monthly by state regulators. The longest series is Nevada’s. According to the UNLV Center for Gaming Research’s compilation of Nevada Gaming Control Board figures (Nevada Sports Betting Totals: 1984-2025, February 2026), Nevada books kept between 5.13 and 7.45 percent of all sports wagers in each year from 2018 to 2025, with football and basketball close to the all-sports figure and parlay cards far above it.

Year	Handle, all sports ($bn)	Win% all sports	Football	Basketball	Baseball	Parlay cards
2018	5.01	6.01%	5.65%	6.60%	4.59%	24.08%
2019	5.32	6.19%	6.55%	5.61%	5.10%	29.84%
2021	8.15	5.46%	5.66%	5.53%	4.77%	32.09%
2022	8.71	5.13%	5.94%	4.17%	4.94%	28.32%
2023	8.26	5.83%	5.84%	5.24%	5.79%	28.11%
2024	7.89	6.11%	5.40%	6.32%	5.13%	14.50%
2025	8.07	7.45%	7.98%	5.88%	7.18%	32.59%

Two things stand out. First, the all-sports hold sits a point or two above the 4.55 percent that a balanced -110 book would earn, which is consistent with Levitt’s finding that books profit from unbalanced action on favourites and with the higher margins charged on props and same-game parlays. Second, the hold is volatile year to year because the book is carrying risk: 2025’s 7.45 percent and 2022’s 5.13 percent were produced by broadly similar margins and very different results. Parlay cards hold 14 to 40 percent because the margin compounds across legs, a mechanism explained in our guide to how parlays work and why they lose.

How a prediction market differs

A sportsbook sets the price and takes the other side of every bet. A prediction market such as Kalshi or Polymarket does neither: the price is set by buyers and sellers in an order book, the exchange charges a fee on trades rather than building a margin into the odds, and the equivalent of the vig is the bid-ask spread plus that fee. A Kalshi contract at 52 cents bid and 54 cents ask has a 2-cent spread, which is 2 percent of the $1 payout, split between two traders rather than paid by one. Whether that is cheaper than -110 depends on the fee schedule and the liquidity of the specific market; our sportsbook versus exchange comparison tool runs the arithmetic for a given price. The legal position of those exchanges differs by state and is tracked in our Kalshi legality tracker.

What this guide does not cover

This guide describes how prices are formed, not how to beat them. It does not cover the specific models bookmakers use, which are proprietary; the practice of limiting or closing accounts of winning bettors, which varies by operator and jurisdiction; promotional pricing such as boosted odds, which deliberately runs below the standard margin; or in-play pricing, where the margin is typically higher and the line changes every few seconds. The Nevada figures are reported hold, which includes the effect of results, promotions and futures bets settled in later periods, so they are not a direct measure of the margin on any single market. Figures are as of November 2026; the UNLV series is updated annually.

Frequently asked questions

How do bookmakers decide the opening line?

An oddsmaker or model estimates each outcome’s probability from power ratings and situational factors, converts it to a price, then adds a margin so the two sides sum to more than 100 percent. The opener is posted with low limits and adjusted as professional bettors react to it.

What is the bookmaker’s margin at -110?

At -110 on both sides, each side implies 52.38 percent and the two add to 104.76 percent. The margin on money wagered is 4.55 percent, and a bettor must win 52.38 percent of equal-sized bets to break even.

Do sportsbooks try to balance the money on each side?

Not as a rule. Steven Levitt’s 2004 study of 19,770 NFL wagers found at least two-thirds of bets on one side in nearly half of games, with the book letting the imbalance stand and profiting from bettors’ preference for favourites, which covered only 48.2 percent of the time from 1980 to 2001.

How much do sportsbooks actually keep?

Nevada books held between 5.13 and 7.45 percent of all sports wagers in each year from 2018 to 2025, according to UNLV’s compilation of Nevada Gaming Control Board data. Parlay cards held 14 to 40 percent.

Why does the line move when there is no news?

Because money from bettors the book considers sharp has arrived. Books move toward the number that splits or stops professional action rather than toward an even split of all bets, and in large markets other books follow within minutes.