Fundamentals
What Is a Prediction Market? A Complete Explainer
What prediction markets are, how they aggregate information into probabilities, their accuracy record vs polls and pundits, and a brief history from the Iowa Electronic Markets to Polymarket.
The one-sentence definition
A prediction market is a financial exchange where traders buy and sell contracts whose payout depends on the outcome of a real-world event, and whose live price is therefore the market's collective probability estimate of that event happening.
How a contract works
The standard prediction-market contract is binary. It pays $1.00 if a specific event occurs ("Will the Fed cut rates in March?") and $0.00 if it doesn't. The contract trades continuously between $0.00 and $1.00, and its current price is the implied probability: a $0.34 quote on YES means the market thinks there's a 34% chance the event happens.
Multi-outcome markets ("Who will win the election?") are simply collections of binary contracts — one per candidate — whose prices sum to ~$1.00.
Why they're more accurate than polls
Three reasons:
- Skin in the game. Survey respondents pay no cost for a wrong answer. Traders pay with capital, which filters out noise and bias.
- Information aggregation. The order book pools the private information of every participant into a single number. No single survey can match this.
- Real-time updating. Prediction markets reprice within seconds of news. Polls take days and average across stale data.
Empirically, the Iowa Electronic Markets beat polling averages in the majority of U.S. presidential cycles since 1988. PredictIt and Polymarket continued the pattern through 2016, 2020, and 2024.
A brief history
Election betting has happened in some form since the 19th century — there were formal curb markets on U.S. presidential races on Wall Street through the 1930s. Modern prediction markets begin with the Iowa Electronic Markets (1988), an academic experiment that became the template. Intrade popularized the format internationally in the 2000s before shutting down in 2013. PredictIt ran a CFTC-approved version inside the U.S. through 2022. Polymarket, launched in 2020 on Polygon, became the dominant venue in 2024 with the U.S. presidential cycle and now hosts the largest prediction-market liquidity ever seen.
What prediction markets are good for
Forecasting elections, corporate decisions, geopolitical outcomes, scientific milestones, and any binary event with a clear resolution criterion. Companies like Google, Microsoft, and Ford have run internal prediction markets to forecast project deadlines and demand.
What they're bad at
- Very low-liquidity questions where price can be moved by a single trader.
- Events with poorly-defined resolution criteria (oracle disputes are real).
- Long-tail events where the market simply hasn't formed an opinion yet.
How to participate
Outside the U.S., the dominant venue is Polymarket. U.S. residents typically use Kalshi. Both run a true central limit order book; both settle in dollars (USDC for Polymarket, USD for Kalshi). For a side-by-side comparison see our Polymarket vs Kalshi guide.
Frequently Asked Questions
How accurate are prediction markets?
Academic studies of the Iowa Electronic Markets, PredictIt, and now Polymarket consistently show prediction markets outperform political polls and expert pundits, especially in the final 30 days before an event. Their forecasts are well-calibrated — outcomes priced at 70% happen ~70% of the time.
What is the difference between a prediction market and gambling?
A prediction market is a financial market on a binary or multi-outcome event whose price is the consensus probability. Sportsbooks set prices internally and earn a vig; prediction markets are peer-to-peer order books where traders set prices. Most regulators in the U.S. classify event contracts as derivatives, not gambling.
Why do prediction markets work?
Two reinforcing mechanisms: skin-in-the-game forces participants to back beliefs with capital (filtering noise), and the order-book aggregates dispersed private information into a single price. The Hayek-style price discovery is what makes them more accurate than surveys.
Who invented prediction markets?
The modern academic prediction market dates to the Iowa Electronic Markets, launched at the University of Iowa in 1988 to forecast U.S. presidential elections. Earlier informal versions of political wagering existed for centuries on London and Wall Street curb markets.
Are prediction markets manipulable?
In theory, yes — anyone can post a large order at an off-market price. In practice, manipulation attempts on liquid markets are rapidly arbitraged away by other traders, and the cost of moving price grows non-linearly with size. Thin markets remain genuinely manipulable.
Put this into practice
Track the top 100 most profitable Polymarket traders in real time, get Telegram alerts on their buys, and run consensus algos against their positions.
Start free trial