What is a prediction market? The complete primer

The academic origins

Prediction markets were born in 1988 in an economics department, not a trading floor. The University of Iowa launched the Iowa Electronic Markets (IEM), a not-for-profit experimental platform where researchers could see whether pricing real contracts on political outcomes would outperform the polls. They got approval from the CFTC to run it as an academic research project.

The result was striking. The IEM consistently predicted US presidential and Senate elections more accurately than the published polls, often by several percentage points. A small group of amateur traders, risking their own money (capped at $500 per person), beat national polling operations with professional weights and sophisticated methodology.

That's the foundational claim of prediction markets: forcing people to put money on the line reveals information that surveys cannot capture.

A survey asks you what you think. A prediction market asks you what you'd bet on. The gap between the two answers turns out to be huge. A large fraction of people polled will say what's socially acceptable, what they vaguely believe, or what they assume the researcher wants to hear. Very few people will put $100 behind an opinion they don't actually hold.

The two ideas that power prediction markets

1. The wisdom of crowds

The wisdom-of-crowds thesis, popularized by James Surowiecki's 2004 book of the same name, holds that large, diverse, independent groups of non-experts often out-predict experts. The reasoning is that each individual has a small piece of information; their errors are distributed unevenly; when you aggregate them, errors cancel and truth emerges.

The classic demonstration is Francis Galton's 1906 observation at a livestock fair: 787 people guessed the weight of an ox. No individual was accurate but the mean of their guesses was within 1% of the true weight.

A prediction market is a mechanism for aggregating the crowd. The price movement is the aggregation happening in real time.

2. Efficient-markets reasoning

In any exchange, the price is set by the marginal buyer and seller. If a contract is trading at 30 cents and you think it's worth 60, you have a direct incentive to buy it — your purchase itself pushes the price up until it approaches your estimate, or until you run out of money. The market equilibrium is the price where marginal buyer and seller disagree by the smallest amount.

A large, liquid prediction market on an event should, by this logic, reflect the best available probability estimate. Any systematic bias would be arbitraged away by profit-seekers.

In practice, prediction markets are not perfectly efficient. They tend to have mild biases toward long-shot events (the favorite-longshot bias), and liquidity is thin on obscure markets. But for high-volume markets — US elections, major Fed decisions, championship sports — the price is a sharp aggregate.

How a single contract works

Consider a market on "Will the S&P 500 close above 7000 on Dec 31, 2026?"

The market has two contracts: YES (settles $1 if true) and NO (settles $1 if false).
YES and NO prices must sum to $1 (by arbitrage — otherwise you could buy both and make free money).
If you think the probability is 40%, you'd buy YES if it's trading below 40¢ and NO if it's trading below 60¢.
At resolution, you get $1 for every winning share you hold.

A key insight: the price is a probability. If YES is at 0.58 and NO is at 0.42, the market is saying the S&P will close above 7000 with 58% probability. This is the most important feature of a prediction market — the numbers it produces are immediately interpretable as probabilities without any further translation.

Compare to a sportsbook: you have to convert American odds (+150, -210) or decimal odds (2.50) into an implied probability, and then strip out the vig. Prediction markets skip that step.

Market types and structure

Binary markets

The most common form. A single YES/NO contract on a clearly-defined event. Example: "Will the Fed cut rates at the March FOMC meeting?"

Resolution is unambiguous once the event happens. Prices move as new information arrives (Fed minutes, economic data, speeches by officials).

Multi-outcome markets

Several mutually-exclusive outcomes. Example: "Who will win the 2028 GOP presidential nomination?" with contracts for each candidate. Each contract is a YES/NO on that candidate specifically, and their YES prices sum to (approximately) 1 across all candidates.

In practice they sum to slightly more than 1 because of spread and the possibility of a write-in winner. The gap is small on liquid markets.

Scalar markets

Rarer and more complex. The contract pays an amount tied to a continuous variable. Example: "Pays $X per dollar based on BTC's closing price on Dec 31." Neither Polymarket nor Kalshi uses pure scalar markets at scale — both split scalar questions into binary bucket ranges ("will BTC close between $100k and $120k").

What makes prediction markets hard

Resolution ambiguity

"Will the US enter a recession in 2026?" sounds clean but it isn't. Which definition of recession? NBER's after-the-fact declaration? Two quarters of negative GDP? As of what measurement source?

Well-designed markets specify resolution criteria precisely: "Resolves YES if the NBER Business Cycle Dating Committee declares a recession starting between Jan 1 2026 and Dec 31 2026, otherwise NO." Ambiguous markets become disputed markets, and disputed resolutions are the main failure mode of the category.

Liquidity

A market only works if there are people on both sides. Thin markets have wide spreads, slow price discovery, and unreliable probability estimates. Polymarket and Kalshi solve this partly with market makers (automated liquidity providers) and partly with volume attraction on flagship markets.

Obscure markets — "Will Mongolia hold an early election in 2027?" — rarely develop deep liquidity, and their prices should be treated with skepticism.

Sophisticated traders vs casual users

Because prediction markets are financial, sophisticated traders dominate niche markets and the casuals get picked off. A sports bettor placing $200 on "Will the Chiefs cover?" is competing with quant traders running live odds models across multiple exchanges.

This isn't unique to prediction markets — the same is true of the stock market — but it's worth knowing before you size up.

The categories prediction markets have mastered

Politics

The breakthrough category. Polymarket's 2024 US election saw $3.6 billion in volume. Kalshi prevailed in October 2024 federal court to offer political event contracts in the US. Markets cover primaries, general elections, cabinet confirmations, international elections (UK, France, Germany), and Supreme Court decisions.

Economics

Kalshi's native territory. Fed rate decisions, CPI prints, unemployment numbers, GDP calls. Macro hedge funds use these as cheap instruments to hedge specific policy scenarios.

Sports

Competitive with traditional sportsbooks. The advantage of prediction markets over sportsbooks is the 0–1% fee vs 4–5% vig. Polymarket covers major leagues, championships, season-long markets. Kalshi is building out sports following its 2024–2025 sports contracts approval.

Crypto

Prices on BTC and ETH milestones, ETF approvals, protocol upgrades, exchange hacks. Useful for crypto-native hedging — a BTC holder can buy "BTC stays above $X" cheaper than equivalent options.

Entertainment

Oscars, Grammys, Emmys, box office. Polymarket dominates. Entertainment markets have sharp insider flow; they're harder to trade profitably than they look.

Science

Nobel prizes, FDA approvals, SpaceX launches, AI benchmark milestones. Thin liquidity but interesting as probability-reading even if you don't trade.

How prediction markets compare to their alternatives

vs Polls

Polls tell you what people say they believe. Prediction markets tell you what people are willing to bet on. The gap matters especially when there's a socially-acceptable answer that conflicts with private belief (Brexit, Trump 2016).

Markets also update continuously. A poll is a snapshot; a market is a live price.

vs Pundits

Pundits are rarely held accountable for their calls. Markets reveal pundits' track records implicitly: over enough events, the market beats most individual commentators.

vs Forecasting models

Statistical models (FiveThirtyEight-style) are transparent and well-methodology. But they can miss what markets catch: qualitative signals, late-breaking news, insider information, shifts in momentum. The best use is both: models as a baseline, markets as the real-time update.

vs Sportsbooks

Sportsbooks take risk against you and charge vig. Prediction markets match you against other traders. For the same event, the sportsbook's "odds" include a 4–5% house margin that the market exchange doesn't charge. Informed bettors prefer exchanges.

vs Options/derivatives

Equity options are scalar derivatives on stock prices. Prediction markets are binary derivatives on events. For specific event risk ("will the Fed cut?" "will this bill pass?"), prediction markets express the question more directly than any options construction.

The big objections

"Isn't this just gambling?"

Legally, yes, in many jurisdictions. Intellectually, prediction markets are closer to derivatives than to sports betting — they express economic risk, provide price discovery for events, and have legitimate hedging use cases. The academic literature treats them as informational mechanisms, not as gambling.

The regulatory distinction depends on whether the contract is "economically purposive" (hedging real risk) or "purely speculative". Kalshi, a CFTC-regulated DCM, argues its contracts are economic instruments. Polymarket is treated as gambling in most jurisdictions that have addressed it directly.

"Doesn't this create perverse incentives?"

The worry is that if you can bet on bad outcomes, you have an incentive to cause them. In theory, yes. In practice, market sizes are tiny relative to the incentive to cause real-world events. A $100M political market doesn't meaningfully change election incentives that run into the tens of billions.

CFTC rules explicitly prohibit markets on certain categories (war, death, terrorism) partly to pre-empt this concern.

"Aren't these easy to manipulate?"

Liquid markets are hard to manipulate for long. You'd have to take a losing position to push the price, and sophisticated traders would immediately arbitrage you. Thin markets are more manipulable but also matter less — their prices are already noisy.

Why this category is having a moment

The three forces pushing prediction markets into the mainstream in 2024–2026:

Regulatory clarity on the US side. Kalshi's October 2024 federal court win opened legal event contracts to US retail. The 2022 CFTC Polymarket settlement became a reference model for future enforcement.
Mainstream media citation. 2024 election coverage cited Polymarket prices in the same paragraphs as polls. This normalized the category for a broad audience.
The crypto stack. Stablecoins, cheap L2 gas, and wallet UX improvements made Polymarket usable by non-crypto-natives. The onramp is still imperfect but has closed ~80% of the gap to centralized exchange UX.

The result: prediction market volume is now orders of magnitude above what it was pre-2024, and the trajectory is still accelerating.

FAQ

Frequently asked questions

Are prediction markets legal?+

It depends on the platform and jurisdiction. In the US, Kalshi is CFTC-regulated and legal in all 50 states; Polymarket blocks US IPs after a 2022 settlement. Globally, legality varies widely — see our country-by-country guide.

How accurate are prediction markets?+

Large, liquid markets on well-defined events (elections, Fed decisions, championships) are consistently more accurate than polls or pundit forecasts. Thin or ambiguous markets are less reliable.

Can you make money on prediction markets?+

Yes, but mostly for traders with an information edge or for small-scale players willing to do real research. Casual betting tends to lose over time, just as in any market. See How to win on Polymarket.

What's the difference between Polymarket and Kalshi?+

Polymarket is decentralized, crypto-settled, globally accessible except to US residents. Kalshi is centralized, USD-settled, federally CFTC-regulated, legal in all 50 states. Full comparison here.

How do prediction markets resolve?+

Each market specifies a resolution source. Polymarket uses UMA's optimistic oracle (which resolves most markets automatically, with dispute escalation if needed). Kalshi resolves internally, with criteria listed on each market page.

Can I hedge real-world risk with prediction markets?+

Yes, this is a legitimate use case. A company could hedge election risk, a hurricane-exposed business could buy weather contracts, a portfolio manager could hedge Fed-decision risk. Whether this hedge is available in sufficient size depends on market liquidity.

Are prediction markets just a crypto fad?+

No. They've existed academically since 1988 (IEM) and commercially since Intrade in the 2000s. The current moment is driven by Polymarket's crypto rails and Kalshi's CFTC approval, both of which solved prior regulatory and UX barriers.

How do I start?+

Read What is Polymarket?, What is Kalshi?, and Polymarket vs Kalshi. Then pick the platform that fits your location and preferences, and sign up via our referral links above.