A prediction market turns a question about the future into a tradable contract: shares that pay out if a specific outcome happens and expire worthless if it doesn't. Because the price of a "yes" share moves with buying and selling, it functions as a running, crowd-sourced probability estimate rather than a single forecast from one analyst or pollster.
Centralized platforms like Kalshi operate as regulated exchanges, matching orders and settling contracts under direct oversight from regulators such as the US Commodity Futures Trading Commission. Crypto-native platforms such as Polymarket instead settle trades on-chain, using stablecoins for collateral and a smart contract to hold funds until a market resolves. Earlier decentralized projects like Augur pushed the model further by having token holders act as the resolution mechanism themselves, staking value on the correct outcome and earning fees for reporting it honestly.
The hardest technical problem is getting real-world results onto the blockchain accurately, which is why decentralized markets depend heavily on oracle systems or staked-reporter disputes to confirm what actually happened. Prediction markets are used well beyond crypto prices, covering elections, sports, and economic data, and research has repeatedly found their aggregated odds track real outcomes at least as well as expert polling. Risks include thin liquidity on niche markets, ambiguous contract wording that complicates settlement, and a patchwork of regulation: some jurisdictions treat these contracts as legitimate financial instruments, while others classify them as gambling and restrict or block access entirely.