An on-chain prediction market is a platform where users buy and sell shares in the outcomes of future events (elections, sports, crypto prices, world events). Shares in the winning outcome pay $1; losing shares pay $0. Prices reflect the market's collective probability estimate. Polymarket is the largest example, processing hundreds of millions in monthly volume on political and world events.
What Is an On-Chain Prediction Market?
3 min read
The short version
A prediction market lets you bet on whether something will happen, but it is structured as a market, not a bookmaker. You buy "yes" shares (this will happen) or "no" shares (it will not). If "yes" shares trade at $0.65, the market thinks there is a 65% chance it happens. If you buy at $0.65 and the event occurs, you earn $1 (35 cents profit). If it does not occur, you lose your $0.65. The crowd's combined bets create a real-time probability estimate that is often more accurate than polls or expert predictions.
How It Works
Structure: each market has two outcomes (binary: yes/no) or multiple outcomes. Each outcome has a share that trades between $0 and $1. All outcome shares for a market sum to $1. When the event resolves, the winning outcome shares pay $1 each; all others pay $0. Resolution: determined by an oracle (a designated data source, often a decentralized oracle like UMA or a trusted committee). On-chain advantage: settlement is automatic and trustless (smart contract pays out based on oracle resolution), markets are permissionless (anyone can create one), and trading is 24/7 without geographic restrictions. Polymarket (on Polygon) is the dominant crypto prediction market, processing $100M+ monthly volume on US elections, geopolitical events, crypto prices, and cultural outcomes. Augur (Ethereum L1, earlier generation), Azuro (sports), and Hedgehog (various) are alternatives.
Trading on a Polymarket election market
Market: "Will Candidate X win the 2024 presidential election?" Current prices: Yes shares: $0.52. No shares: $0.48. You believe Candidate X has a higher than 52% chance, so you buy 1,000 "Yes" shares at $0.52 each ($520 total, paid in USDC). The election occurs. Candidate X wins. Your 1,000 Yes shares each pay $1 = $1,000 total. Profit: $1,000 - $520 = $480 (92% return). If Candidate X had lost: your shares pay $0. Loss: $520 (100% of position). The market price leading up to the election fluctuates as new information arrives (polls, events, news). You could sell your Yes shares before resolution if the price rises (taking profit without waiting for the event) or cut losses if it drops.
What People Get Wrong
Prediction markets are just gambling
They produce something gambling does not: an aggregated probability signal. The market price is a real-time consensus forecast that consistently outperforms polls and expert panels for binary events. The information aggregation function has genuine value for decision-making, forecasting, and research. But yes, individual participation is economically equivalent to a bet.
Market prices are always accurate predictions
Prices reflect the market's best estimate given current information and participation. They can be wrong (low liquidity markets, manipulation, herding). They tend to be well-calibrated over many events (events priced at 70% happen roughly 70% of the time) but any single market can misprice.
On-chain prediction markets are legal everywhere
Regulatory status varies enormously. The US CFTC regulates prediction markets (Kalshi has a CFTC-regulated platform for event contracts). Polymarket blocked US users after a CFTC settlement. Other jurisdictions have different stances. Accessing these markets from restricted jurisdictions may violate local laws.
Keep Reading
Sources & Further Reading
- Polymarket
The largest on-chain prediction market for world events
- UMA Optimistic Oracle
Documentation for the oracle system resolving Polymarket outcomes
Questions People Also Ask
- How are prediction markets resolved?
- By oracles: designated data sources that report the outcome to the smart contract. Polymarket uses UMA (optimistic oracle where anyone can dispute resolutions). Some markets use multiple data sources with majority agreement. The resolution mechanism is the most critical trust assumption in any prediction market.
- Can prediction markets be manipulated?
- In theory: someone can buy shares to move the price and signal a false probability. In practice: manipulation is expensive (you are buying shares at a loss if the market reverts) and corrected by informed traders who see mispricing as profit opportunity. Large, liquid markets are harder to manipulate than thin ones.
- How accurate are prediction markets?
- Studies show well-functioning prediction markets are among the most accurate forecasting tools available, consistently outperforming polls for elections and expert panels for other events. Calibration is typically good: events priced at 80% happen about 80% of the time. But they are not infallible, especially for unprecedented or low-information events.