What Are Prediction Markets? How They Work, and What They Do Well

Prediction markets are exchanges where people buy and sell contracts tied to a future event. A contract usually pays a fixed amount, often $1, if an outcome happens and nothing if it does not. Because the payout is fixed, the price behaves like a probability: a Yes contract trading at 62 cents implies roughly a 62% chance.

That is the mechanism in one sentence. The important part is what sits underneath it: an order book, collateral, resolution rules and a settlement process. A prediction market is not a sportsbook with a different interface. In most modern venues you are trading against other users, not against a house that sets a price and keeps the margin.

A contract is a pair of claims

Most prediction markets use binary contracts. Every Yes share is matched by a No share, and the pair is backed by $1 of collateral. If the event happens, Yes redeems for $1 and No for $0. If it does not, the reverse. Polymarket describes this as a fully backed Yes/No pair: $1 of collateral creates one of each, and the winning token is redeemable for $1.Polymarket's documentation

The price therefore has a natural ceiling and floor. A Yes share at 30 cents costs $30 for 100 shares. If the event happens, those shares redeem for $100, a profit of $70 before fees. If it does not, the $30 is gone. Per share, the maximum loss is the purchase price and the maximum gain is $1 minus that price.

You do not have to hold to resolution. You can sell before the event if the price moves, which is how most active traders manage risk. That ability to exit is a structural feature of an exchange, not a courtesy.

Prices come from an order book

Polymarket uses a central limit order book. Buyers post bids, sellers post asks, and the best bid and best ask define the market. The displayed price is usually the midpoint. Polymarket's order book documentation notes one exception: if the spread is wider than $0.10, the venue shows the last traded price instead.

Suppose the best bid for Yes is 34 cents and the best ask is 40 cents. The midpoint is 37 cents, or 37%. A buyer pays 40 cents, not 37. A seller receives 34 cents. The 6-cent gap is the spread, and it is a real cost.

New markets discover prices through complementary orders. One trader bids 60 cents for Yes; another bids 40 cents for No. Because 60 + 40 = 100 cents, the orders can match: $1 of collateral creates a Yes share for the first buyer and a No share for the second. The first trade sets a reference point, and subsequent orders move it.

Who provides liquidity

Liquidity comes from market makers, algorithmic traders, and ordinary users who leave limit orders on the book. A market maker quotes both a bid and an ask, earns the spread when both sides fill, and carries inventory risk while waiting. That risk is the reason liquidity is not free: the maker is paid for standing ready to trade when others want to exit.

The practical point for a trader is that "the price" is not a public statistic. It is the best available offer at that moment, for that size. A market can look liquid at the top of the book and be thin one level deeper.

Settlement decides the payout

A market's title describes the question; its resolution rules define the answer. Rules specify the resolution source, the end date and how edge cases are handled. Most Polymarket prediction markets use the UMA Optimistic Oracle. Anyone can propose an outcome with a bond, typically $750. There is a two-hour challenge period. A dispute triggers a second proposal; a second dispute escalates to UMA's Data Verification Mechanism, where token holders vote. An undisputed market resolves in about two hours, while a disputed one can take four to six days.Resolution rules

That process matters before you trade, not after. A contract that asks whether something will be "announced" may resolve on an official statement rather than a press report. A contract with a date may use a specific time zone. If the rules are ambiguous, the market price can be right about the world and still pay the wrong side.

What they are good at, and where they fail

Good at

Prediction markets aggregate dispersed information. People with different models, data and incentives buy and sell until the price reflects a consensus. A classic survey of the literature by Justin Wolfers and Eric Zitzewitz found that market-generated forecasts are typically fairly accurate and outperform most moderately sophisticated benchmarks.Journal of Economic Perspectives

They also update quickly. New polling, a court filing or a shipment report can be priced within minutes if the market is deep enough. That speed is the main reason prediction markets are useful to journalists, analysts and traders who need a running estimate rather than a once-a-week model.

The other underrated use is hedging. The CFTC gives the example of a citrus farmer buying a weather contract to offset the risk of a freeze. A trader can hedge an exposure without finding someone who wants the exact opposite bet; the exchange matches the two sides.CFTC explainer

Where they fail

Thin markets are the obvious weakness. A market with a few hundred dollars of depth can be moved by a modest order, and the resulting price may say more about the order book than about the event. A large field experiment across 817 markets found that price shocks were still visible 60 days later, although the effects faded and were smaller in markets with more traders, more volume and an external probability benchmark.Rasooly and Rozzi

Resolution ambiguity is the second weakness. Oracles and dispute processes reduce it, but they do not eliminate it. A badly written rule can turn a forecasting trade into a legal interpretation trade.

The third is the price itself. A cheap contract is not automatically a bargain: the lower the price, the larger the relative payoff but the lower the chance of winning, and it is easy to overestimate the edge. That is a reason to demand a larger margin of safety when the price is far from 50 cents, not a reason to avoid every longshot.

Finally, prediction markets are not frictionless. Capital is locked until resolution unless you sell, fees and spreads consume edge, and access depends on jurisdiction. The CFTC treats event contracts as derivatives and regulates designated contract markets; the rules are not the same as sports betting, and they are still being litigated in the United States.CFTC explainer

None of this makes prediction markets useless. It makes them a tool with a specific failure mode: they are excellent at aggregating information when the market is deep and the question is clean, and poor at both when it is not.

The short version

A prediction market turns a probability into a tradable price. The price comes from an order book, the payout comes from a resolution rule, and the quality of the signal depends on liquidity and contract design. If you remember one thing, make it this: the number on the screen is an offer, not a forecast, and the rules are part of the trade.