Markets vs. Sportsbooks

A prediction market looks familiar to anyone who has used a sportsbook. Both attach numbers to upcoming events, both move those numbers as new information arrives, and both reward a correct read on the future. The resemblance sits on top of a structural difference that changes nearly everything underneath. A sportsbook takes the other side of every wager it accepts, while a prediction market matches participants against each other and takes no side at all.

Who stands on the other side of a trade turns out to explain most of what feels unfamiliar in practice: where the numbers come from, what participation costs, how winning accounts are treated, and what exiting early means. This guide works through the comparison one mechanism at a time and translates sportsbook vocabulary into trading vocabulary along the way. It assumes familiarity with sportsbooks and no trading background at all.

What is the difference between a prediction market and a sportsbook?

A sportsbook is a counterparty. The operator sets the odds, accepts wagers against its own account, and earns its margin from the difference between the odds it offers and the probabilities it believes are true. A prediction market is a venue. It runs an order book where participants buy and sell event contracts with each other, and the operator earns trading fees while holding no position in any outcome. An event contract is a standardized agreement tied to a yes-or-no question about a future event. It settles at a fixed value, typically $1, if the stated outcome occurs, and at $0 if it does not (CFTC).

That single change, from trading against the operator to trading through the operator, drives every row of the comparison.

Both sides of the industry describe their own models plainly. Exchange operators explain that they earn trading fees rather than trading against their users (How does Kalshi make money?), and the question comes up often enough that venues answer it by name (Is Polymarket the house?). Reporting on the category draws the same line between the two structures, contrasting a business that profits when its customers lose with one that earns a fee for standing between a buyer and a seller (ESPN).

The house model carries a genuine service inside it. Because the operator takes the other side of every wager, a book can quote a price on demand for anything in its catalog, at any hour it operates, in any size within its limits. The margin in the odds is the fee for that immediacy. An order book asks participants to supply that liquidity to one another instead, which is why a market price comes with a bid, an ask, and sometimes a gap between them. Neither arrangement is a defect; they are different answers to the question of who provides the other side of a trade, and the rest of this guide traces what each answer implies.

How do prediction market prices compare to sportsbook odds?

A sportsbook expresses its view in an odds format: American, decimal, or fractional. Every format encodes the same two quantities, an implied probability and the operator's margin. Part of learning to read a line is learning to separate them. A prediction market expresses the consensus as a single number, the contract price, which reads directly as a probability. A contract trading at 34¢ implies roughly a 34% chance, because the contract settles at $1 when the outcome occurs and at $0 when it does not. Economists have long treated these prices as probability forecasts that aggregate dispersed information (Wolfers and Zitzewitz).

Converting American odds into probabilities makes the difference concrete. With L as the posted line:

$$p_{\text{favorite}} = \frac{|L|}{|L| + 100} \qquad\qquad p_{\text{underdog}} = \frac{100}{L + 100}$$

In plain language: for a negative line, divide the line's absolute value by itself plus 100; for a positive line, divide 100 by the line plus 100. Suppose a matchup is quoted at -150 on the favorite and +130 on the underdog. The favorite's implied probability is 150 divided by 250, or 60%. The underdog's is 100 divided by 230, or about 43.5%. The two numbers describe the same event, yet they sum to 103.5%. The extra 3.5 points are the vig.

A prediction market's two sides are complements by construction. A Yes contract at 58¢ corresponds to a No contract at 42¢, because exactly one of them will settle at $1. In a live order book the quotes on the two sides can drift a cent or so apart, and that gap is the spread, an artifact of resting orders rather than a designed margin. The displayed price itself typically comes straight from the book, often the midpoint of the best bid and best offer (How are prices calculated?).

Costs have not disappeared on the market side; they have moved into the open. Crossing the spread costs the difference between the bid and the ask, and venues charge trading fees published on their own schedules (Kalshi, Polymarket). Whether a given trade costs less at a book or on a market depends on the spread, the fee schedule, and the line in question. The structural difference is that a market itemizes its costs while a book embeds them, which can make market pricing easier to reason about and to compare. How prices map to probabilities, including the effect of spreads on that reading, is the subject of Prices and Probabilities.

Who sets the odds if there is no bookmaker?

At a sportsbook, a trading team sets the opening line, drawing on models, power ratings, and the prices already posted elsewhere. From that point the operator adjusts the line for two reasons: to track new information, and to manage its own exposure when too much money lands on one side. The number you see is a quote from one firm, published on its schedule and moved at its discretion.

On a prediction market, nobody sets the price. It emerges from the orders participants submit. One trader offers to buy Yes at up to 58¢, another offers to sell at 60¢, and the venue's matching engine pairs orders whenever a buyer and a seller agree on a level. The market price is simply the level of the most recent agreement. When news breaks, the price moves because traders cancel, reprice, and resubmit orders, not because an operator decided the line should move.

The two loops differ in one step. The sportsbook loop runs through the operator's judgment, while the market loop runs through whoever shows up to trade. Order Books covers the mechanics behind the second loop: bids, asks, depth, and matching.

The model is older than event contracts. Betting exchanges have run person-to-person order books on sporting events since the early 2000s, and their educational material teaches the same core motions of posting a price and trading out of a position (Betfair's exchange education hub). Prediction markets inherit that architecture and extend it beyond sports.

For someone arriving from a sportsbook, the practical difference is agency. At a book, the posted number is take-it-or-leave-it. On a market there is a third option: place an order at your own price, inside the spread or away from it, and wait for the market to come to you. If you think the number is wrong, you are not limited to accepting someone else's quote. You can become the quote. The price everyone then sees reflects every participant who made that same choice, which is why markets with many active traders tend to produce sharper prices than quiet ones.

Do prediction markets limit winning traders?

Sportsbooks manage risk partly by managing customers. Every wager is against the house account, so a customer who wins consistently is a recurring cost, and operators respond by reducing maximum stakes or declining the action altogether. The practice is documented and openly defended. At a Massachusetts Gaming Commission roundtable, major US operators described limiting a small share of accounts as what allows them to keep quoting competitive lines to everyone else (ESPN). Whatever position you take on the practice, the logic behind it is structural. A firm that quotes prices to the public with its own money at stake protects itself from the people who beat those prices.

An exchange does not face that decision, because it has no exposure to protect. A consistent winner on an order book costs the venue nothing. The winner's fees are revenue like anyone else's, and the winner's gains come from the traders who took the other side, not from the operator. There is no structural incentive to remove skilled participants. Venues maintain rulebooks, position limits, and market-integrity rules, but these are generally written to apply uniformly rather than to single out profitable accounts.

The absence of curation cuts both ways. On an exchange, nobody screens the other side of your trade. The order that fills yours may come from a casual participant, a professional, or an automated strategy, and you will rarely know which. One independent analysis of several years of exchange trade data found that aggressive orders crossing the spread underperformed on average, while the patient resting orders on the other side captured the difference (The Microstructure of Wealth Transfer in Prediction Markets). Nobody will stop you from winning on a prediction market, and nobody will protect you from participants who are better informed. Both follow from the same design.

Can you cash out on a prediction market?

Cash out, as sportsbooks use the term, is a repurchase offer. The operator computes what an open wager is currently worth under its own model, applies a margin, and offers to settle early at that number. The offer is a product feature: it can be withdrawn during volatile stretches, it is available only where the operator chooses to enable it, and its price is whatever the operator says it is.

A prediction market does not need a cash-out feature, because exiting is simply selling. A position is a contract you own, and you can sell it to any willing buyer at the market price at any point before resolution, the same way you would sell a share of stock. Venue documentation states this directly (Can I sell early?). Exchange traders have long had their own names for the motion: trading out, or greening up when the exit locks in a gain across every outcome (Betfair exchange guides).

A worked example shows the shape of it. Suppose you buy Yes at 34¢ and a strong report lands the next morning, repricing the market to 62¢. Selling closes the position with a 28¢ gain per contract before fees, without waiting for the event and without anyone's permission. The same motion manages the downside. If the market instead moves against you and trades at 20¢, selling recovers 20¢ per contract rather than risking the slide toward zero. Between entry and resolution, the position is an asset with a live price, and the full menu a stock trader works from is available: taking profits, cutting losses, and scaling in and out.

The honest caveat is liquidity. A sale needs a buyer, and the price you can actually realize is the current bid rather than the last trade printed on the chart. In an active market the two sit close together; in a thin one the bid can sit several cents lower, and a large exit can push through the visible depth. A cash-out button always produces a number, because the operator prices it. A market exit is priced by the other participants, which tends to serve you well in liquid markets and demands more patience in quiet ones.

What are parlays, lines, and cash outs called on a prediction market?

Most sportsbook concepts have a direct counterpart in trading vocabulary. The table maps the common terms, and the third column records what changes in translation.

Three rows repay a closer look.

Parlays translate the least cleanly. A sportsbook parlay is a single wager whose legs must all succeed, priced from the operator's payout table. On a prediction market, the equivalent exposure is assembled from positions across several markets, and a venue can also list a packaged multi-leg contract as a market of its own. In both forms, the fair combined probability is the product of the legs only when the legs are independent. Correlated legs make that multiplication misleading, on a market just as much as on a slip.

Handle and volume look interchangeable and are not. Handle counts the money staked. Market volume is typically notional, counting each contract at its full settlement value rather than at the cash that changed hands, so a one-dollar position in contracts priced at a cent can register as a hundred dollars of volume. Analysts who track the category caution against reading the two figures as the same quantity (Next Event Horizon).

The push row hides the largest behavioral change. A graded slip can push and refund; an event contract resolves Yes or No according to criteria written before trading opens, including which sources count and how edge cases are handled (resolution documentation). Sportsbook experience builds no equivalent habit of reading rules before acting, so the final section treats it as a skill of its own.

Which sports betting skills carry over to prediction markets?

A great deal transfers, because the underlying job is unchanged: form a probability, compare it to a price, and act only when the two disagree.

Probability thinking transfers whole. Converting a number into an implied probability, devigging a pair of lines, and asking whether you would take either side of a price are the same mental motions as reading a contract at 34¢. Anyone who has shopped for the best line has also practiced the market version of price comparison, since the same outcome can trade at different prices on different venues.

Benchmarking against the close transfers as a discipline. In sharp sports-betting circles, decisions are graded on closing line value, the gap between the price taken and the final pre-event price, because beating the close has been shown to track long-run results (Pinnacle's CLV explainer; Buchdahl's closing-line efficiency analysis). The same self-audit works on contracts. Watch whether markets tend to move toward your entries or away from them. A trader whose entries sit consistently on the right side of later prices is likely doing something repeatable; one whose entries sit consistently on the wrong side is being told so by the market.

Bias awareness transfers with new evidence. The favorite-longshot bias, the long-documented tendency for longshots to be overpriced relative to how often they occur, appears in event-contract data as well. One analysis of several years of exchange trades found the cheapest contracts returned far less than their prices implied (Becker). A cheap contract is not automatically cheap value, on either kind of venue.

Capital discipline arrives intact. What a sportsbook customer calls bankroll management, a trader calls sizing positions against capital. The vocabulary changes; the arithmetic and the temperament do not.

Three assumptions do not survive the move.

The counterparty assumption. Beating a sportsbook means beating one firm's posted number. On a market, a mispricing has to be bought from another participant, and quick corrections may come from professionals or automated strategies. The skill of spotting a stale number survives the move; the assumption that the other side is slow does not.

The grading assumption. A slip is graded against the result everyone watched. A contract settles against written resolution criteria, and edge cases are decided by the text. Reading a market's rules before trading is the one habit with no sportsbook equivalent, and the few minutes it takes are repaid more reliably than anything else in this guide.

The catalog assumption. A sportsbook's catalog is sports. The account you use to trade a championship market also prices elections, economic data releases, awards shows, and questions with no league behind them at all. The instrument, a contract that settles at $1 or $0 against written criteria, is the same everywhere. What differs from venue to venue is structure and access, which is the subject of Venue Types.

The comparison reduces to a single sentence: a sportsbook sells you a price, while a prediction market lets you trade one. Everything else in this guide, from the itemized costs and the tolerance for winners to the exit that needs no permission and the rules that decide settlement, follows from that sentence. The guides below pick up the mechanics where this one leaves off.