Skip to content
Prediction Markets101

Oracle

An oracle is the mechanism that brings real-world information on-chain to resolve a prediction market. Polymarket uses UMA's optimistic oracle; Kalshi uses centralized human resolution.

Prediction Markets 101 editorial team Updated April 16, 2026 2 min read

Definition

An oracle is the system that tells a prediction market what actually happened when a market resolves. In a world where markets live on-chain but outcomes happen off-chain (elections, sports, weather), something has to bridge the two. The oracle is that bridge.

Plain-English explanation

When Polymarket lists the market "Will the Fed cut rates in March?", eventually the FOMC releases a statement. Someone — or some system — has to read that statement and say: "Yes, they cut" or "No, they didn't". The market then pays YES shares at $1 or NO shares at $1 based on that answer. The oracle is whatever provides that answer.

How Polymarket's oracle works

Polymarket uses UMA's optimistic oracle. The flow:

  1. Someone proposes an answer after the event happens. They post a bond.
  2. There's a dispute window (typically 2 hours).
  3. If nobody disputes, the proposed answer becomes final.
  4. If disputed, the question escalates to UMA token holders who vote.

"Optimistic" because most answers pass without dispute. Proposers who lie lose their bond; honest disputers earn rewards.

How Kalshi's oracle works

Kalshi uses centralized human resolution. Kalshi staff reads the named resolution source at the named time and publishes the result. Users have a 48-hour dispute window to flag issues; final decisions happen under CFTC oversight.

Centralized is simpler and faster but relies on trust in Kalshi. Decentralized is more cryptographically robust but slower in edge cases.

Why the oracle design matters

The oracle is the single most critical design choice in a prediction market. A bad oracle — slow, disputable, or captureable — can cause markets to pay out incorrectly, which destroys the platform's credibility.

Both UMA (for Polymarket) and Kalshi's internal process have strong track records. Neither has had a platform-wide resolution failure. But both have had individual controversial resolutions.

See also