Sponsored
← Education

How prediction markets work: prices, probabilities and payouts

A beginner’s guide to prediction market contracts, prices and payouts, with a worked example, resolution rules and limits to market forecasts.

June 3, 2026·
How prediction markets work: prices, probabilities and payouts

A prediction market is a market for contracts whose payouts depend on defined future outcomes. Participants buy and sell positions on questions such as whether an economic release will exceed a threshold or a team will win a match. Prices provide a changing view of how traders value those outcomes.

In a standard binary contract, the winning side pays $1 and the losing side pays $0. A price of $0.62 is therefore commonly read as a 62% market-implied probability. It is an estimate drawn from trading, not a measurement of the event’s true probability.

Sources checked: 22 October 2026. The examples below use fully paid, $1-payout binary contracts. Other contract designs, collateral arrangements and exceptional payouts require their own rules. The CFTC’s event-contract introduction describes both binary contracts and other payout structures.

Jump to the worked example, price interpretation, order book, Yes/No relationship or reading checklist.

Follow a $1 contract from purchase to payout

Consider a hypothetical market asking whether a named economic release will exceed a stated threshold. Its rules identify the publication, release date and measurement used to decide the result.

You buy 100 Yes contracts at an executed price of $0.62. The purchase costs $62 before fees. What happens next determines your return:

OutcomeAmount receivedGain or loss before fees
Yes wins and each contract pays $1$100$38 gain
No wins and the Yes contracts pay $0$0$62 loss
You sell all 100 contracts at $0.70 before resolution$70$8 gain

The sale example assumes a buyer accepts all 100 contracts at that price. It is not a guaranteed exit. The $100 winning payout includes the recovery of the original $62 purchase cost; it is not $100 of profit.

These are hypothetical calculations. Check the contract’s actual payout and any fees before applying them to a live market.

What does a prediction-market percentage mean?

For a standard $1-payout contract, multiplying the price by 100 gives the familiar implied percentage: $0.25 corresponds to 25%, $0.62 to 62% and $0.90 to 90%. This convention makes prices useful for discussing uncertainty.

It does not mean 62% of participants voted Yes. A market price results from orders and executed trades, not a one-person-one-vote survey. Nor does a 62% price establish that the outcome will occur in exactly 62 out of the next 100 cases.

Justin Wolfers and Eric Zitzewitz’s research on interpreting market prices identifies conditions under which prices reflect average beliefs. It also explains how risk preferences and the distribution of beliefs affect the relationship. The useful interpretation is an estimate with assumptions, rather than an exact identity between price and truth.

Election markets add another common confusion. A candidate’s expected vote share and probability of winning are different quantities. A close vote-share forecast can coexist with a much higher or lower win probability, depending on uncertainty and the electoral rules. Research on communicating election forecasts explains that distinction. Comparing a polling percentage directly with a market’s win probability does not establish which forecast is better.

How buying, selling and price changes work

In an order-book market, bids are offers to buy and asks are offers to sell. The gap between the best bid and ask is the spread. The amount offered at each price tells you how much can trade there. Kalshi’s order-book guide shows both prices and quantities.

The displayed percentage may differ from either executable price. For example, a $0.60 best bid and $0.64 best ask have a midpoint of $0.62. Buying from the available seller costs $0.64; selling to the available buyer receives $0.60, subject to quantity and fees. Polymarket’s pricing documentation describes its midpoint display and use of the last traded price when the spread is wide.

A limit order sets the price you are willing to accept. It can remain unfilled or fill only partly. An immediately executable order takes available liquidity and can trade across several prices if its size exceeds the quantity at the best quote. Kalshi’s limit-order guidance explains the price-control distinction.

New evidence can change participants’ orders and move the price. Trading flows and changes in liquidity can move it too. A price jump alone does not identify which news item caused it, or prove that someone has superior information.

Our World Cup liquidity analysis shows why a large trading-volume figure is different from the depth available for a new order.

Why Yes and No are linked, but quoted prices can differ

For the ordinary all-or-nothing binary contract, one Yes position and one No position together pay $1 at settlement. On Polymarket, its positions documentation describes complete sets backed by one unit of collateral and the splitting and merging of those sets.

That relationship creates incentives to keep prices consistent. It does not require every pair of visible buy or sell quotes to total exactly $1. Quotes include spreads, may refer to different sides of an order book, and must be available in the required quantity.

For example, two hypothetical asking prices of $0.63 for Yes and $0.40 for No total $1.03. Buying one of each would cost more than the ordinary $1 combined payout, before fees. The payout relationship has not disappeared; the quoted purchase costs differ from the payout.

Any apparent arbitrage needs executable orders, applicable fees and the platform’s collateral and redemption rules. The relationship alone does not make every displayed price efficient or guarantee a cost-free trade.

How markets combine information, and where they fall short

Participants can bring different evidence, models and expectations to the same question. Someone who judges a contract underpriced may buy it, while someone who considers the price too high may sell. Trading gives those disagreements a price.

Financial incentives can encourage careful research because an incorrect position can lose money. They do not ensure that every participant is informed or that mistaken prices immediately correct. Correcting a price takes capital, a willingness to accept risk and enough confidence in the contract’s eventual resolution.

When reading a market, consider:

  • Depth and spreads. A small available order book can make a displayed price costly to trade or easy to move.
  • Shared assumptions. Many participants can rely on the same mistaken source or model.
  • Participation and incentives. The people able and willing to trade need not represent the population, and their objectives may differ.
  • Rules and evidence. A contract can concern a narrower condition than its headline suggests.
  • Market conduct. Financial stakes do not make prices immune to manipulation or other misconduct.

Read our behavioural-biases guide for favourite–longshot bias and herding. Market prices can be useful inputs alongside other evidence; they should not be presented as automatically superior to every expert or forecasting method.

What happens when the event ends?

The market’s rules determine the evidence needed to establish the outcome. Trading close, outcome determination and payout can happen at different times. An apparent result in a news report may not yet satisfy the contract’s named source or deadline.

Kalshi’s market FAQs explain that the exchange may wait for official data before determining and settling a result. Polymarket’s resolution documentation describes UMA-based resolution and a separate Chainlink TWAP route for Up/Down markets. Those processes should not be treated as interchangeable.

Our resolution and settlement guide covers deadlines, evidence, dispute routes and exceptional outcomes. A dispute or delay needs to be assessed against the relevant rules.

How to judge a prediction market’s accuracy

A forecaster is calibrated when events given a particular probability occur at roughly that frequency over many appropriate cases. Among a sufficiently large set of forecasts near 70%, a calibrated method would see the event happen about 70% of the time.

A single 70% forecast that loses is not enough to establish poor calibration. Equally, selecting the eventual winner once does not prove that the probability assigned beforehand was accurate. Record the forecast time, question and eventual outcome, and compare like with like across many cases.

Election results can also be correlated: several state forecasts in the same election are not necessarily independent tests. The election-forecasting research linked above explains why limited data and correlated errors complicate evaluation. Accuracy claims need a stated dataset, time horizon and comparison method.

Read a market before deciding whether to trade

You can begin by inspecting public market pages on Polymarket or Kalshi, without placing an order. For a question you understand, check:

  1. The exact condition that pays, including its date and named evidence source.
  2. Whether the displayed percentage is a midpoint, last trade or executable quote.
  3. The bid, ask and quantity available at the prices you would use.
  4. The fees and the difference between a potential payout and profit.
  5. The determination process, eligibility requirements and funding route.

Continue with our prediction markets versus sports betting comparison or the wallets, deposits and withdrawals guide. For jurisdiction-specific sources, use the regulatory tracker and policy references.

Sponsored
Sponsored