How prediction markets work
At their core, prediction markets let participants trade real money on whether a specific event will happen (yes) or not (no) via event contracts, and the market price of these contracts reflects the crowd’s collective belief about the likelihood of the event.
Each contract is a binary contract, and has a payout rule attached and specified in advance – for instance, “Pays $1 if candidate X wins the election, $0 otherwise.”
The best markets are those with clear and concise rules, terms, and resolution criteria, free from ambiguity amongst traders.
Contracts are typically priced between 1¢ and 99¢, using one-cent increments. These prices easily convert to implied probabilities: a contract trading at 75¢ implies a 75% chance of the event happening.
Some platforms go even more granular, offering fractional pricing (e.g., 1.7¢, 3.5¢ and so on). This format provides more precise pricing and forecasts but is not as user-friendly to casual traders.
Traders can place buy or sell orders on these contracts, as prediction markets use the law of supply and demand via its order book to dynamically set prices in real time: the more people want to buy a contract, the more its price will rise.
For example, if a well-known expert publicly says “I think there’s a 90% chance candidate X wins,” but the market is sitting at 80%, opportunistic traders will rush to buy at the bargain price of $0.80. The increased demand can push the price up. If buyers keep buying, the price will keep climbing.
Prediction Markets Education
Are prediction markets accurate?
Michigan University economist Justin Wolfers may have put it best when he said, “The promise of prediction markets isn’t that they’re perfect, it’s that everything else is worse.”
Prediction markets have attracted attention for their accuracy because, when conditions are right, they often outperform traditional forecasting methods. Their greatest strength is harnessing the “wisdom of crowds,” the idea that aggregating many independent guesses can produce remarkably accurate predictions. These markets tap into dispersed knowledge: each trader may know a little bit (a local rumor, an analytical model, a gut feeling, etc.), and the market price combines all those bits together into one number.
As long as traders have diverse information sources and incentives to be honest (i.e. to make money), the market’s collective forecast tends to be very hard to beat. In many case studies, prediction market prices have been shown to equal or beat the accuracy of polls, expert opinions, and statistical models.
For example, election markets have a strong track record: the Iowa Electronic Markets famously out-predicted many polls in U.S. presidential races, and prediction market prices can even be used to improve polling forecasts by accounting for late-breaking information.
In one analysis, combining market odds with poll results lowered prediction errors compared to using polls alone. Most recently, in the 2024 U.S. presidential election, prediction markets Polymarket and Kalshi both correctly made Donald Trump the betting favorite ahead of election day in face of polling and pundits that argued otherwise.
Not only have these markets correctly forecasted elections, but they’ve been applied in fields from business to public health. Additionally, prediction markets can be used as hedging tools to mitigate risk.
