How prediction market
contracts work

The Lifecycle of a Contract

Every prediction market contract follows a defined path from listing to payout. Understanding each stage removes hidden costs and prevents avoidable mistakes.

Stage 1: Creation

Prediction platforms organize their products in a hierarchy: Events (broad topics like "2026 US Midterms") contain Markets (specific tradeable questions like "Will Republicans win the House?") which contain Outcomes (the actual YES/NO shares you buy and sell).

At creation, each market is given resolution criteria — the precise rules that define how it settles. These are the most important words in any market listing and the first thing any systematic trader should read. Ambiguous resolution criteria have caused millions of dollars in disputed payouts.

Stage 2: Trading

Once live, contracts function as binary options trading between $0.01 and $0.99. The price floats continuously based on order flow. Makers post limit orders that sit on the book waiting to be filled; Takers execute market orders that immediately fill against existing resting orders.

This is where the first structural cost appears. Every time a Taker executes a market order, they pay the platform's taker fee — a percentage of the contract value that compounds dramatically over time. On Kalshi, this peaks at roughly 3.5¢ per contract at 50/50 odds.

Stage 3: Resolution

After the event occurs, the exchange consults its predefined resolution source to determine the outcome. Kalshi uses official government data — NWS Daily Climate Reports for weather, BLS for inflation, FOMC statements for interest rates. Polymarket uses the UMA Optimistic Oracle, a decentralized truth-verification system involving proposers, bonds, and challenge periods.

The gap between resolution sources matters more than most traders realize. Two platforms listing the "same" market can resolve differently based on which data source they trust.

Stage 4: Settlement

The contract resolves to a binary value:

• If the event was TRUE: YES contracts pay $1.00, NO contracts pay $0.00
• If the event was FALSE: NO contracts pay $1.00, YES contracts pay $0.00

Your profit on a resolved YES contract = $1.00 − your purchase price. Buy YES at 30¢, event resolves YES: profit is 70¢ per contract. Buy YES at 30¢, event resolves NO: loss is 30¢ per contract.

Price = Implied Probability (Always)

This is the foundational mechanical truth of prediction markets: the price of a contract is mathematically identical to the market's implied probability of that event occurring.

This relationship has a critical implication: your edge in prediction markets is not the ability to predict outcomes — it is the ability to identify when the market's implied probability is wrong. You don't need to know what will happen. You need to know when the crowd has mispriced what will happen.

MECE Market Sets and the Free Donut

Some markets on Kalshi are structured as Mutually Exclusive and Collectively Exhaustive (MECE) sets — groups of outcomes where exactly one must be true and only one can be true. Temperature bracket markets are a common example: "High temp will be below 60°F / 60–70°F / 70–80°F / above 80°F."

The pricing property of MECE sets is mathematically guaranteed: the sum of all contract prices must equal exactly $1.00. If they don't, a risk-free arbitrage exists.

Resolution Criteria: Where Disputes Are Born

Resolution criteria are the strict rules defining how a market settles. In practice, vague language in these rules has caused major disputes across platforms.

Real examples from the research record:

• The Government Shutdown Case (2024): Polymarket resolved a government shutdown market as YES. Kalshi resolved the same event as NO. Arbitrageurs who held positions on both platforms lost money on both sides simultaneously — an outcome mathematically impossible if resolution criteria had been identical.
• The Venezuela "Invasion" Dispute: Conflicts over whether a specific military action qualified as an "invasion" under each platform's definition created litigation-level disagreements.
• NYC Weather: Kalshi resolves New York City temperature at Central Park (KNYC). Polymarket uses LaGuardia Airport (KLGA). These stations diverge by 1–5°F daily — making true cross-platform arbitrage on NYC weather markets structurally impossible.

The rule: Before placing any trade, read the resolution criteria. If you can construct a scenario where the outcome is ambiguous under those criteria, don't trade the market.

The Hidden Costs: Spreads and Slippage

The Bid-Ask Spread

The bid-ask spread is the gap between the best price buyers are willing to pay (bid) and the best price sellers are willing to accept (ask). On a liquid market this might be 49¢ bid / 51¢ ask — a 2¢ spread. On a thin weather market it might be 82¢ bid / 88¢ ask — a 6¢ spread that represents a meaningful percentage of the contract's value.

When you execute a market order, you immediately lose half the spread by definition. You buy at the ask (higher) or sell at the bid (lower). This is a cost that appears before the event even begins, and it is separate from the taker fee.

Slippage

Slippage is what happens when your order is large enough to consume multiple price levels in the order book. In deep markets (politics, major economic events), slippage is negligible on typical retail position sizes. In thin markets — particularly Kalshi weather contracts — the research documents order books with only 10–20 contracts available at the best ask.

When Markets Are Cancelled: Resolved N/A

Occasionally an event cannot be determined, a market is cancelled before resolution, or the underlying event is changed materially. Platforms handle this differently:

• Kalshi: Has codified rules for event halts. In most cancellation scenarios, the market settles at the last traded price, returning a prorated amount to holders. If a primary subject of a market dies before resolution (e.g., a candidate market), Kalshi may invoke specific contract halt provisions.
• Polymarket: Disputed resolutions go through the UMA Optimistic Oracle's challenge period. A proposer submits a resolution with a bond (typically $750 USDC). If no one challenges within 2 hours, the market settles. In some high-stakes disputes, the process has been vulnerable to whale manipulation — in March 2025, a single holder controlling 25% of UMA voting power manipulated a $7M market resolution.