Prediction markets and sportsbooks price outcomes in different ways

A prediction market is an exchange on which traders buy and sell contracts tied to a future event, rather than placing a bet with an operator. Each contract settles at either zero or 100% once the outcome is known, and in the meantime its price moves between roughly one cent and a dollar depending on how the market rates the chance of that outcome.

The standard structure is a pair of contracts, one for "Yes" and one for "No". If the Yes contract trades at $0.93, the market is implying a 93% chance the event happens; the No contract will sit at $0.07, implying a 7% chance of the opposite. That price is not set by a company quoting a line. It emerges from a central limit order book, where a trade only happens when one participant is willing to buy at a price another is willing to sell at, according to Wikipedia.

The underlying idea, drawn from crowdsourcing theory, is that a market made up of many people with money at stake can estimate a probability more reliably than any single forecaster. Eric Zitzewitz, an economics professor at Dartmouth, has described it this way: "Financial markets are generally pretty efficient, and the evidence suggests that the same is true of prediction markets. There's no virtue-signaling in an anonymous market when you're betting."

James Surowiecki, cited in the same entry, sets out three conditions for this kind of collective judgement to work: the participants need diverse information, they need to decide independently of one another, and the organisation producing the price needs to be decentralised rather than controlled by a single authority.

The record is mixed. Studies have found a bias in which prices for events further in the future drift towards 50%, attributed to traders' reluctance to lock up money for a long period. Prediction markets also misjudged both the 2016 UK referendum on EU membership and the 2016 US presidential election, with researchers pointing to traders anchoring on early odds and reinforcing that view rather than updating on new information.

Organised betting on predicted outcomes is not new. Wikipedia's entry notes a wager on the 1503 papal succession, already described at the time as an old practice, and records of election betting on Wall Street dating to 1884; researchers Paul Rhode and Koleman Strumpf have estimated that betting turnover on a US presidential election has run to more than 50% of campaign spending.

How a sportsbook differs

A sportsbook works on a different principle. The bookmaker acts as the price-setter and market maker, quoting odds on an outcome and accepting the bet upfront, rather than matching one customer's view against another's in an open book. Those odds build in a margin for the operator, commonly described as the vigorish, so that the bookmaker holds an edge over the customer regardless of which side wins.

That edge has been estimated at roughly an 11-10 advantage for the bookmaker, narrowing to around 6-5 on smaller wagers, according to Wikipedia's entry on sports betting. Odds are quoted in decimal, fractional or moneyline formats depending on the market, but all three express the same thing: an implied probability that already has the operator's margin folded in.

The practical distinction, then, is about where the price comes from. A prediction market's price is the output of traders betting against each other with no house edge embedded in the mechanism itself, while a sportsbook's odds are set by the operator and structured so it profits over time regardless of the result.

Regulators generally treat the two similarly in practice. Prediction markets are considered a form of gambling by many governments and are banned in some jurisdictions, and some researchers have reported that they carry the same potential for compulsive use as other betting products, Wikipedia's entry states.