How prediction markets actually resolve

A prediction market is only as good as its resolution. If the answer to "did this happen or not" is unclear, ambiguous, or disputed, the market's final price is meaningless. So the resolution mechanism — who decides, how, and how disputes get handled — is one of the most important things to understand before trading seriously.

Kalshi: internal resolution committee

Kalshi resolves markets via a designated team that monitors official sources defined in the contract specification. Each market has a "Source of Resolution" listed in the rules — for example, "the Bureau of Labor Statistics nonfarm payrolls release" for an economic market.

The resolution process is fast (usually within hours of the source publishing) and deterministic if the source is unambiguous. The downside is that you're trusting Kalshi as a centralized arbiter. In practice they have a strong incentive to resolve correctly — their CFTC license depends on it — but you don't have crowd-based dispute mechanics the way crypto-native platforms do.

Polymarket: UMA optimistic oracle

Polymarket resolves through UMA's optimistic oracle. The flow:

• When a market closes, anyone can propose an answer (Yes, No, or 50/50 if ambiguous) along with a USDC bond.
• A challenge window opens — typically 2 hours. During that window, anyone can dispute the proposed answer by posting a bond of their own.
• If unchallenged, the proposed answer is final and the market settles.
• If challenged, the dispute goes to UMA's tokenholder vote (DVM — Data Verification Mechanism). Holders of the UMA token vote on the correct answer over a 48-hour window. The losing side's bond is forfeited.

The system is "optimistic" because it assumes most resolutions are uncontroversial (true) and only escalates the disputed ones. In practice over 99% of Polymarket resolutions go through unchallenged.

When markets are genuinely ambiguous

Both platforms have a "50/50 split" or "invalid" outcome for cases where the resolution criteria can't be applied. If you bought Yes and the market resolves 50/50, you get back 50 cents per share — better than zero, worse than $1.

Markets become ambiguous more often than you'd expect:

• The source of resolution stops publishing data on time.
• The event is delayed past the contract's resolution date.
• Multiple official sources give different answers.
• The wording leaves real ambiguity about what counts (e.g. "by mid-2026" — does July 1 count?).

How to read resolution rules

For any market you plan to put real money on, the rules section is more important than the question. Three things to check every time:

• The source. Where does the answer come from? Is it a single official publication? Multiple sources reconciled? An oracle's judgment call?
• The deadline. What happens if the underlying event hasn't occurred by the resolution date? Some markets resolve No automatically; others extend; others resolve invalid.
• The edge cases. What if the source is unavailable? What if the event is partial (e.g., a candidate withdraws after winning)? Good rules anticipate these; bad rules leave you arguing on Discord.

Both platforms have strong track records

For 99% of markets, resolution is unambiguous and fast. The dispute machinery exists for the rare 1% case. If you're not trading near edge cases, this whole topic mostly doesn't matter to you. It only matters when it matters — and at that point, you want to have understood it before, not after.

What to read next

• Where to look for edge in prediction markets
• Common mistakes new traders make