What Is a Prediction Market? How the Prices Work
Prediction markets let people trade contracts tied to real-world events. Here is how they work, why they exist, and what the prices actually tell you.
A prediction market is a marketplace where people buy and sell contracts whose payout depends on the outcome of a future event, such as an election, a central bank rate decision, or whether a product ships on time. Each contract typically pays a fixed amount (often $1) if the event happens and nothing if it does not, so the trading price, somewhere between 0 and 1, reads as the crowd's estimated probability of that outcome. A contract trading at 62 cents implies the market thinks there is roughly a 62 percent chance the event occurs.
The counter-intuitive part is that a few thousand people betting their own money often forecast an election more accurately than a professional pollster surveying the same population. Prediction markets are built around that finding, and this piece explains why it holds, who runs the major markets today, whether they are legal, and the specific conditions under which the crowd gets it badly wrong.
Key takeaways
- A prediction market converts opinions about the future into tradable prices, and those prices behave as real-time probability estimates.
- The core idea is old: pool many people's money and information, and the resulting price tends to forecast better than most individual pundits.
- Two operators dominate today: Kalshi, a US exchange regulated by the CFTC, and Polymarket, a blockchain-based platform that settles trades in cryptocurrency.
- Legal status varies sharply by jurisdiction and platform type, and it remains actively contested.
- Prices beat polls under clean conditions, but thin liquidity, long time horizons, and gameable settlement can each make the number unreliable.
Why prediction markets exist
Before formal prediction markets, if you wanted a numeric read on an uncertain event you relied on polls, expert forecasts, or bookmakers. Each has a weakness. Polls capture stated intentions, not the intensity of belief or the willingness to back it with money. Pundits face no cost for being wrong. Bookmakers set odds to guarantee their own profit, which distorts the number.
Prediction markets solve a specific problem: they aggregate scattered information into a single price that updates continuously and that people have a financial reason to get right. If you know something the market has not priced in, you can profit by trading on it, and the act of trading moves the price toward reality. Economists call this mechanism the wisdom of crowds. The Iowa Electronic Markets, run by the University of Iowa since 1988, showed that markets of modestly sized bets on elections often matched or beat polling.
The simplest analogy is a friendly bet on an election. You and a friend disagree, so you each put money down, and whoever is right collects. A prediction market is that same bet scaled up to thousands of strangers, with a live price that summarizes what all of them collectively believe. The difference from a casino or a sportsbook is that the price itself is meant to be useful information, not just a wager.
Who runs the major markets today
Two operators account for most of the attention and volume, and they represent two very different models.
Kalshi is a US-based exchange that lists event contracts under the oversight of the Commodity Futures Trading Commission (CFTC), the federal regulator for derivatives markets. Users fund accounts in US dollars, trade through a regulated venue, and the exchange resolves outcomes under published rules. Kalshi's appeal is that it puts event trading on a regulated footing that institutions and cautious retail users can engage with. Its constraint is the same: it operates inside a rulebook and a market menu that regulators can expand or narrow.
Polymarket settles trades on a public blockchain, a shared, tamper-evident ledger, using a stablecoin (a cryptocurrency designed to hold a steady value, usually one US dollar). It drew heavy volume during recent US elections and became a widely quoted probability source in newsrooms. Its appeal is scale, global reach, and full transparency, since every trade is visible on the ledger. That same visibility invited scrutiny. Bloomberg has cited Allium data in reporting on possible settlement manipulation on Polymarket, a reminder that the number on the screen is only as trustworthy as the process that decides who was right.
The practical upshot: journalists now quote prediction-market odds alongside polls, traders watch them for signal, and regulators are deciding how far these markets can expand. If you follow markets or policy at all, these prices are already shaping the conversation.
How a prediction market works, step by step
- A market is created. Someone proposes a clearly defined question with a fixed resolution date and rules, for example, "Will the Federal Reserve cut rates at its next meeting?"
- Contracts are listed. The platform issues "Yes" and "No" contracts. A matched pair is worth $1 at settlement, because exactly one of them will pay out.
- People trade. Buyers and sellers set prices through their orders. If "Yes" trades at 70 cents, the market is pricing a 70 percent chance. Prices move as new information arrives, much like a stock price reacting to news.
- The event resolves. After the real-world outcome is known, the market settles. Winning contracts pay $1 each; losing contracts expire worthless.
- Settlement is verified. Someone or something must confirm what actually happened. On regulated exchanges, the operator resolves under published rules. On blockchain platforms, resolution often relies on an oracle, a mechanism that feeds real-world outcome data onto the blockchain, sometimes with a human dispute period.
That final step is the pressure point. If the resolution source is ambiguous, gameable, or contested, the payout can be wrong even when the trading was honest.
Reading a price as a probability, with the math
The single most useful skill is converting a market price into an implied probability and a payout. The rule is simple: for a contract that pays $1 on a "Yes" outcome, the price in dollars is the implied probability directly. A price of $0.62 means a 62 percent implied chance.
Here is a worked example on a concrete contract. Say a "Yes" contract on "Will the Fed cut rates at its next meeting?" trades at $0.62, and you buy 100 of them.
| Quantity | Price per contract | Cost to enter | Payout if Yes | Net profit if Yes | Net loss if No |
|---|---|---|---|---|---|
| 100 | $0.62 | $62.00 | $100.00 | +$38.00 | -$62.00 |
The implied probability is the price itself: 62 percent. To sanity-check that the price is fair to you, compare your own estimate. If you believe the true chance is 70 percent, the market is underpricing "Yes" and the trade has positive expected value: (0.70 x $38) minus (0.30 x $62) equals $26.60 minus $18.60, or about +$8.00 expected per 100 contracts. If you think the true chance is only 55 percent, the same trade has a negative expected value and you would sit out or trade "No".
The "No" side mirrors this. A "No" contract costs $1.00 minus the "Yes" price, so $0.38 here, and pays $1 if the event does not happen. A matched Yes plus No pair always costs $1.00, which is why the two prices sum to a dollar in an efficient market. Any gap between them is a signal of fees, thin liquidity, or stale quotes.
Regulated exchanges versus blockchain platforms
Not all prediction markets are the same, and the differences carry real consequences for legality, access, and trust. The table below lays out the two dominant models in concrete terms.
| Feature | Regulated event exchange (e.g. Kalshi) | Blockchain platform (e.g. Polymarket) |
|---|---|---|
| Oversight | Operates under a financial regulator such as the CFTC | Runs on public blockchain code; regulatory status varies and is contested |
| Money used | US dollars through a regulated broker-style account | Stablecoins held in a self-custody wallet |
| Who can access | Typically verified users in permitted jurisdictions | Global, subject to platform terms and geoblocking |
| How trades settle | Central operator under published rules | Smart contracts plus an oracle or dispute process |
| Data transparency | Exchange-published records | Every trade visible on the public ledger |
Why the prices often beat polls, and when they do not
The headline benefit is a live, quantified forecast. Instead of "analysts are divided," you get a number that moves in real time and that people are financially motivated to correct. A poll captures what people say three days ago. A market price captures what traders will bet on right now, and it absorbs new information the moment it breaks, because a mispricing is money left on the table. Pundits face no cost for being wrong; a trader who is persistently wrong loses capital and stops moving the price.
That accuracy is conditional, not automatic. A market price is a probability, not a prophecy: a contract at 80 cents will still be wrong one time in five. Four conditions in particular degrade the signal.
- Thin liquidity. A price is only meaningful if enough money stands behind it. A low-volume market can be moved cheaply by a single trader, so the number is noisy and easy to distort.
- Long time horizons. Markets resolving years out attract few informed traders and tie up capital for a long time, which weakens the incentive to correct a mispricing.
- Correlated traders. When participants share the same sources and biases, the crowd stops being independent and the price reflects a consensus that can be collectively wrong, the same failure that undermines a poll.
- Manipulation around resolution. Because payouts hinge on how the outcome is decided, a market with vague wording or a small pool of participants can be gamed at settlement, as covered in how settlement manipulation bends prediction markets.
Volume and liquidity are therefore part of the signal, not a footnote to it, which is why analysts study them closely.
Reading a blockchain market's data cleanly
On a blockchain platform like Polymarket, every trade is recorded on a public ledger, which sounds like it should make analysis easy. It does not. A raw trade record arrives as a smart-contract event with hashed addresses, token identifiers, and encoded amounts, with no label for which market it belongs to, which outcome was bought, what the USD value was, or whether the address is a distinct human. To answer a basic question like "how much real money stood behind this market, and did a handful of wallets move the price," each raw event has to resolve to consistent fields: market, outcome, trader, contract quantity, price, USD value, and trade direction. Those fields also have to line up with the same fields from other chains and other venues before any comparison is possible.
Allium normalizes those raw onchain records into standardized datasets across 150+ blockchains, delivered through databases, APIs, and data streams, and maintains prediction market datasets built for exactly this kind of question. Allium is a read layer, not a venue, exchange, broker, or custodian, and it does not offer investment advice. Allium data has been cited by Bloomberg in reporting on prediction markets, including US-based activity on Polymarket during a period of restricted access.
Risks and open questions
Legal status is unsettled. In the United States, the CFTC has allowed some event contracts while contesting others, and court decisions have shifted the boundary. State regulators have separately challenged sports-related event contracts. The line between a legitimate hedging instrument and unlawful gambling is genuinely blurry and still being drawn, as we lay out in whether prediction markets are legal.
Settlement can be manipulated. Because payouts hinge on how an outcome is resolved, bad actors may target the resolution process itself, especially in markets with vague wording or a small pool of participants. This is the subject of active reporting and scrutiny.
Insider information is a live question. When someone trades on private knowledge of an outcome, the ethics and legality are unresolved, and the boundary differs from securities markets, a tension explored in when prediction market trading becomes insider trading.
Cross-border access is inconsistent. Some platforms are blocked in certain countries, and users sometimes circumvent those blocks, which creates legal exposure the platform's headline numbers do not reveal.
They are not universal forecasters. Markets are strong where information is public and outcomes are clean, and weaker for rare, fuzzy, or far-off events where few people can trade with any edge.
None of this is investment or legal advice. It is a description of an evolving landscape, and anyone participating should consult the current rules in their own jurisdiction.
Frequently asked questions
What is a prediction market in simple terms?
It is a marketplace where you trade contracts tied to a future event. A contract usually pays $1 if the event happens and nothing if it does not, so the price between 0 and 1 works like a live probability. A contract at 55 cents implies about a 55 percent chance of that outcome.
How do you convert a prediction market price into a probability?
For a contract that pays $1 on a Yes outcome, the price in dollars is the implied probability directly. A price of $0.62 means a 62 percent implied chance. The matching No contract costs $1.00 minus the Yes price, so $0.38, and the two prices sum to a dollar in an efficient market.
Are prediction markets legal?
It depends on the platform and jurisdiction. In the United States, some event contracts operate under CFTC oversight, while others have been challenged in court and by state regulators. Many blockchain-based platforms restrict access in certain countries. The legal boundary is still being contested, so check the current rules where you live.
Who runs the biggest prediction markets?
Two operators dominate today. Kalshi is a US exchange regulated by the CFTC where users trade event contracts in US dollars. Polymarket is a blockchain-based platform that settles trades in a stablecoin and drew heavy volume during recent US elections.
Why are prediction market prices often more accurate than polls?
Participants back their views with real money, so they have an incentive to be accurate, and anyone who spots a mispricing can profit by correcting it. That tends to produce forecasts that match or beat polls and pundits, but only under clean conditions. Thin liquidity, long time horizons, correlated traders, and gameable settlement can each make the price unreliable.
How is a prediction market different from sports betting?
Mechanically they overlap, since both involve wagering on outcomes. The key difference is that prediction markets are built so the price itself serves as a forecast, prices move continuously as traders react to news, and payouts are typically fixed at $1 per winning contract rather than set by a bookmaker's odds.
What is settlement and why does it matter?
Settlement is the process of confirming what actually happened and paying out winning contracts. On regulated exchanges the operator resolves under published rules; on blockchain platforms resolution often relies on an oracle and a dispute period. It matters because a flawed or gameable settlement can produce the wrong payout even when trading was honest, which is where most disputes concentrate.
