A contract, not a bet
A prediction market sells a contract that pays a fixed amount if a stated event happens and nothing if it does not. The standard unit is a contract that settles at one dollar on "yes" and zero on "no". If the market is trading at 63 cents, the buyer of a yes contract pays 63 cents and receives a dollar if the event occurs, a gain of 37 cents; the seller of that contract receives 63 cents and pays a dollar if it occurs, a loss of 37 cents. Nobody in that exchange is the house. The two sides are two customers, and the venue that matched them takes a fee.
That structure is the whole subject. Everything else in this course, the pricing, the legal argument, the integrity risks, the reason bookmakers are nervous, follows from the fact that a prediction market is an exchange on which customers trade a binary contract with each other, rather than a counterparty that sets a price and takes the other side.
The phrase "event contract" is the regulatory name for the same thing. In the United States, the Commodity Exchange Act provides for designated exchanges to list contracts based on the occurrence of an event, while letting the CFTC bar those that involve gaming, unlawful activity, terrorism, assassination or war, and the American venues at the centre of the market since 2024 are exchanges designated under that act. In the rest of the world the same product may be called a binary option, a prediction contract or, by most gambling regulators, a bet.
Price is probability
The defining feature of a binary contract that settles at one dollar is that its price is a probability. A yes contract trading at 63 cents is a market saying there is a 63 per cent chance the event happens; the CFTC describes the price as reflecting traders' perceived probability of the outcome. That is not a metaphor. If a trader believes the true probability is 70 per cent, buying at 63 has positive expected value; if they believe it is 55, selling has positive expected value; and the price settles where the marginal buyer and seller disagree by less than the cost of trading.
This is why the product is called a prediction market rather than a betting market. The output is a forecast, continuously updated, produced by people risking money on it. Economists and political scientists have studied that output since the Iowa Presidential Stock Market, now the Iowa Electronic Markets, opened at the University of Iowa in 1988. The academic case, which long predates the commercial one, is that a market can aggregate dispersed information into forecasts that are typically fairly accurate and outperform most moderately sophisticated benchmarks, because it rewards being right and punishes being wrong in a way that opinion surveys do not.
For a gambling professional the same fact has a more practical meaning. A sportsbook's odds also encode a probability, but with a margin built in: the implied probabilities of every outcome in a market sum to more than 100 per cent, and the excess is the bookmaker's expected profit. On an exchange the yes price and the no price of the same contract sum to one dollar, less the spread. There is no overround, because there is no house to earn it.
Yes and no are the same contract
A convention that confuses newcomers: buying "no" at 37 cents is exactly the same position as selling "yes" at 63 cents. Both pay one dollar if the event does not occur. Venues display both sides as things you can buy because retail customers understand buying better than short selling, but the order book is one book: a bid for yes at a price is the same order as an offer of no at one dollar minus that price. A limit order to buy yes at 63 is matched against a limit order to buy no at 37, and the venue holds the combined dollar in collateral until settlement.
That collateral point matters. Because each matched pair of orders is fully funded, one dollar of collateral per yes and no pair, the exchange never carries credit risk on a settled outcome. There is no margin call, no counterparty default and no liability for the venue to manage. The venue's risk is operational: resolving the event correctly, keeping the book fair and staying on the right side of the law.
Two ways to run the book
Central limit order book. Traders post bids and offers at prices they choose; the exchange matches them. This is how regulated exchanges work and how the largest venues operate. Liquidity comes from market makers, firms paid, often through fee rebates, to keep two-sided quotes in the book so that a customer can always trade at some price. Spreads are tight where volume is high and wide where it is not, and a thin book on an obscure event can show a spread of ten cents or more, which can be a worse price than a bookmaker would offer.
Automated market maker. Some venues, particularly those built on public blockchains, instead pool liquidity and let a formula set the price as trades shift the pool's balance. The customer trades against the pool rather than another customer. The design removes the need for market makers but exposes liquidity providers to loss when the price moves, and it produces its own manipulation risks, discussed in lesson four. The largest crypto-native venue, Polymarket, published automated market maker contracts, since archived, and now runs a central limit order book, which suggests which design scales.
Where the venues sit
Three kinds of venue exist, and the distinction is legal before it is technical.
Regulated exchanges. In the United States, a designated contract market is an exchange designated by the Commodity Futures Trading Commission and bound by the core principles of the Commodity Exchange Act. The event contracts it lists are federally regulated derivatives, its customers are identified, its funds are held in dollars, and it is subject to the exchange rules on manipulation, position limits and reporting. This is the model that won the argument over election contracts: the CFTC disapproved Kalshi's congressional control contracts in 2023, a federal court set that order aside in September 2024, and the CFTC dropped its appeal in May 2025. It is also the model that carried sports contracts across the United States from January 2025, an expansion the federal appeals courts are divided on: in 2026 the Third Circuit sided with Kalshi against New Jersey and the Ninth Circuit sided with Nevada against Kalshi, as did the Sixth Circuit with Ohio and Tennessee in September.
Crypto-native venues. Contracts are traded in stablecoins on a public blockchain, positions are held in user wallets rather than by the venue, and resolution is performed by an oracle rather than an exchange committee. The largest of them, Polymarket, settled CFTC charges in 2022 that it had run an unregistered event market, paying a $1.4 million penalty and agreeing to wind down non-compliant markets; these venues then served customers outside the United States, often without a local gambling licence, and gambling regulators have moved against them: in February 2026 the Dutch regulator ordered Polymarket to stop offering its markets in the Netherlands. In July 2025 Polymarket acquired a CFTC-designated exchange and clearing house for $112 million in order to re-enter the American market on the regulated model, and the CFTC has since amended that exchange's designation to let brokers carry customer accounts on it.
Licensed betting exchanges. In Britain and a handful of other markets, exchange betting on sport is licensed as gambling: the Gambling Act 2005 treats a person who facilitates bets between others as a betting intermediary providing facilities for betting, and General Betting Duty applies to the commission an exchange charges. Structurally it is the same product, customers trading a binary outcome with each other, but it is regulated as betting, taxed as betting and marketed as betting. The existence of this model is the strongest argument gambling regulators have that a prediction market is not a new thing.
Why the industry cares
A licensed sportsbook in a regulated American state pays a state gaming tax and operates under the state's regulatory scheme, funds responsible gambling programmes, in most states restricts customers to those 21 or over, and geofences to the state that licensed it. A designated exchange offering a contract on the same game holds no state gaming licence, so pays no state gaming tax, typically accepts customers from 18, and takes the position, shared by the CFTC, that federal law lets it operate in all fifty states without a state licence. Several states dispute that in court. Where the two coexist, the exchange is offering the same wager on different terms, and the terms are better for the customer and worse for the state.
That is the commercial fact underneath every headline. It is also why the largest American sportsbooks moved into the product in 2025: DraftKings agreed in October to acquire a CFTC-designated exchange, and FanDuel launched event contracts with CME Group in December, offering sports contracts only in states where online sports betting is not yet legal. The product is not going away, and the question this course exists to answer is what a professional in the gambling industry should understand about it.
What this course covers
Lesson two explains how prices form, who provides liquidity and what the fees are, because the economics of an exchange are unlike a book. Lesson three is the legal structure: the Commodity Exchange Act, the CFTC, the state gambling laws and the argument between them, which at the time of writing has divided the federal appeals courts and is being litigated in many states at once. Lesson four covers resolution, manipulation and insider trading, the integrity risks that are specific to this product. Lesson five puts the product beside a sportsbook and asks what a customer actually gets. Lesson six is distribution: brokerages, apps, media and partnerships. Lesson seven is a decision framework for an operator, a supplier or a regulator who has to respond.
A closing note on vocabulary. This course uses "prediction market" for the venue and "event contract" for the product, because those are the neutral terms. Whether an event contract on a football game is a bet is the question the courts are deciding, and a course that assumed the answer would be worth less than one that explains the argument.