Get early access

Markets4 min read

How prediction markets work, and what their odds really mean

A prediction market contract pays $1 if an event happens, so 30¢ means about 30%. How prices form, how Kalshi and Polymarket work, and when odds mislead.

By Kristian FagerlieFounder & CEO, Sikt Intelligence

A prediction market is a place where people trade contracts on yes-or-no questions about the future. A contract typically pays $1 if the event happens and $0 if it does not. So if a contract trades at 30 cents, the market is putting the event at roughly 30%.

That simple mechanic turns scattered opinions into a single, constantly updated number. Here is how it works, what the price really means, and when to be careful with it.

How a prediction market works

  1. A question is listed with clear rules: what counts as "yes", which source decides, and by when. For example: Will the central bank cut rates at its next meeting?
  2. Traders buy "Yes" or "No" contracts. A Yes and a No for the same question together always pay out exactly $1.
  3. The price moves with supply and demand. If more people want Yes, the Yes price rises. News, polls and data push it around.
  4. The question resolves. Winning contracts pay $1; losing contracts pay nothing.

Because traders profit from correcting a wrong price, the price tends to move toward what the best-informed people believe.

What the odds really mean

A price of 30¢ is best read as "about 30%, according to the people willing to bet on it." Three things make it an approximation rather than a pure probability:

  • Fees and the bid–ask spread. The price to buy and the price to sell differ, so the "true" market probability sits somewhere between them.
  • The cost of waiting. Money tied up in a contract that resolves in two years could earn interest elsewhere, which can push long-dated prices away from the true odds.
  • Who is trading. A market reflects the views, and the enthusiasm, of its participants.

Kalshi and Polymarket: the two big ones

The two best-known prediction markets work on the same principle but grew up differently:

  • Kalshi received approval from the US Commodity Futures Trading Commission (CFTC) in November 2020 as a designated contract market, the first regulated US exchange built for event contracts (CFTC). It trades in dollars.
  • Polymarket began as a crypto-based market and operated outside the US from 2022. In July 2025 it bought QCEX, a CFTC-licensed exchange, for $112 million to return to the US market, and in September 2025 the CFTC cleared the way for that return (Axios, CoinDesk).

Both list questions on politics, economics, technology, culture and more. Rules and availability vary by country.

Are prediction markets accurate?

Often, yes, but not uniformly. A large study of Kalshi covering over 300,000 contract prices from 2021 to 2025 found that prices are informative and become more accurate as markets approach closing (Bürgi, Deng and Whelan, 2026).

The same study found a clear favorite-longshot bias: cheap contracts on unlikely outcomes win far less often than their price implies. Buyers of contracts under 10¢ lost over 60% of their money on average. We explain why in the favorite-longshot bias.

Research on who drives accuracy also finds that it tends to come from a relatively small number of skilled traders (Yale Insights).

When to be careful with market odds

  • Thin markets. On quiet questions, a few trades can move the price a long way.
  • Long shots. Very low prices tend to overstate the odds.
  • Long horizons. Prices on events years away are distorted by the cost of waiting.
  • Fine print. A market resolves on its written rules, which can differ from the headline question.
  • Breaking news. Prices can overshoot and then correct.

Markets and AI forecasts: two lenses

A market price shows what traders are willing to bet. An evidence-based AI forecast shows what checked facts and base rates support. They fail in different ways, which is why looking at both is more informative than looking at either.

That is the idea behind the Sikt feed: every question shows Sikt's probability next to the market's, and when the two disagree, the gap is worth a closer look. See it in the predictions feed. The odds shown on our site are illustrative, and nothing here is investment advice.

Key takeaways

  • A prediction market contract pays $1 if the event happens, so a 30¢ price means roughly 30%.
  • Prices move with supply and demand, and tend to move toward the best-informed view.
  • Kalshi has been CFTC-regulated since 2020; Polymarket returned to the US through a CFTC-licensed exchange in 2025.
  • Market odds are informative but biased at the extremes: long shots tend to be overpriced.
  • Read market prices alongside evidence-based forecasts, not instead of them.

FAQ

How do prediction market odds work?

Each contract pays $1 if the event happens. Its price in cents is roughly the market's probability: a 65¢ contract means about 65%.

What is the difference between Kalshi and Polymarket?

Both are prediction markets. Kalshi has been a CFTC-regulated US exchange since 2020 and trades in dollars. Polymarket started as a crypto-based market outside the US and returned to the US in 2025 through a CFTC-licensed exchange it acquired.

Are prediction markets better than polls?

They are often at least as accurate, because traders combine polls with other information and have money at stake. They are not reliable everywhere: thin markets and long shots can be badly mispriced.

Sources

Keep reading