A prediction market is a trading venue where the price of a contract reflects the market's estimated probability of a future event. Traders buy and sell outcome shares that pay $1 if the event resolves true and $0 if it resolves false. The price between $0 and $1 becomes the crowd's probability estimate, updated continuously as new information arrives.
​

Combined monthly volume across the largest prediction venues rose from under $5 billion in September 2025 to roughly $24 billion in April 2026 (Pew Research, May 2026). That growth has pulled prediction markets out of an academic-curiosity bucket and into the same volume tier as mid-cap derivatives exchanges. This article describes the mechanism, not the legal status of any particular venue.
​

Prediction markets price events by selling binary outcome shares. A "Yes" share on an event pays $1 if the event happens; the matching "No" share pays $1 if it does not. Because exactly one share pays out, Yes and No prices sum to roughly $1 at all times. The Yes price, read directly, is the market's implied probability that the event happens.
​

Two pricing mechanisms dominate. The first is a continuous double auction (CDA) running on an order book. Traders post bids and asks for Yes and No shares; matching engines execute at the best price. This is the venue model behind Kalshi, Limitless on Base, and most CFTC-regulated event exchanges. CDA venues externalize liquidity risk to market makers, who can quote tighter spreads when volume is high.
​

The second is an automated market maker. Early prediction markets relied on Robin Hanson's Logarithmic Market Scoring Rule (LMSR), a cost-function maker that always quotes a price and bounds the operator's worst-case loss. LMSR sacrifices some price precision for guaranteed liquidity and is the model Augur used at launch in 2015. Modern onchain venues like Polymarket use a hybrid: a CLOB-style order book for execution, with conditional-token primitives borrowed from the AMM tradition for the underlying share representation.
​

Either way, the operator lens is the same. A trader who believes the true probability is higher than the current price buys Yes shares; a trader who believes it is lower sells Yes (or buys No). Profitable disagreement updates the price. The aggregation is mechanical: the marginal price is the marginal trader's belief, weighted by how much capital they put behind it.
​

Onchain prediction markets settle through a smart contract that pays $1.00 in stablecoins per share of the winning outcome and zero per share of the losing outcome. The contract needs a trustworthy answer to the resolution question, which is the hard part. Most onchain venues delegate that answer to an external oracle protocol rather than trying to resolve outcomes themselves.
​

Polymarket, the largest onchain venue, uses UMA's Optimistic Oracle for resolution. After a market's end date, a proposer asserts the outcome onchain. A two-hour challenge window opens. If no one disputes the assertion, the smart contract accepts it as truth and winning shares become redeemable for $1.00 USDC each. If a dispute is filed, the resolution escalates to UMA's Data Verification Mechanism, where UMA tokenholders research and vote on the correct answer. For most uncontested markets, the entire end-to-end flow from event resolution to redeemable USDC takes two to four hours.
​

Limitless, an order-book venue on Base, takes a different approach for its hourly BTC and ETH price markets: it reads Pyth oracle feeds directly to settle short-duration price contracts. That removes the dispute window but only works for outcomes a price feed can answer. Event markets that need human judgment (election outcomes, Supreme Court decisions, "will X tweet Y by Z") still benefit from optimistic-oracle escalation paths.
​

Prediction venues split along two operational axes: where the matching engine and settlement live (onchain stablecoin smart contracts vs centralized exchange infrastructure), and which regulator, if any, oversees the venue. The split is descriptive only. Whether any specific venue is permitted in any specific jurisdiction is a question for the venue's own published terms and the user's own legal review.
​

The table below compares the operational differences between the venue types most readers will encounter.
​

Both venue types host event contracts. The mechanism difference matters most when an outcome is contested or when a trader wants to hold positions across multiple venues. Onchain shares can be moved to another wallet or used in DeFi; CFTC-regulated positions live inside one broker's books.
​

Prediction markets matter because they convert a question into a price, and a price is a forecast that updates in real time. Decades of academic work on the Iowa Electronic Markets and replication-prediction studies show that market-aggregated forecasts outperform most surveys and individual experts, particularly on questions with clear resolution criteria.
​

The aggregation thesis comes from research on information markets: each trader's order is a weighted vote on the outcome, weighted by capital at risk. Traders who hold incorrect beliefs lose money and trade smaller in the next round; traders with better information are funded by their own profits. Over time the marginal price converges toward the marginal informed view. A 2021 study found prediction markets called the replicability of social-science studies with around 73% accuracy, beating expert surveys on the same questions.
​

For builders, the second-order effect is more interesting. Once an outcome has a continuously priced probability, other contracts can reference it. Insurance products, conditional payouts, parametric hedges, and treasury rules can all read a prediction-market price as an input. A DAO that wants to delay a treasury move until "ETF approval probability is above 80%" can read the relevant Polymarket contract directly. The market is becoming a public oracle for soft truths the way Chainlink is an oracle for hard prices.
​

Four venues cover most of the visible activity. Each operates differently enough that grouping them under one label obscures more than it clarifies. Volume figures below are point-in-time and refresh weekly on The Block's prediction-market dashboard.
​

Polymarket runs on Polygon with USDC settlement and a UMA-resolved CLOB. International Polymarket cleared roughly $9 billion in monthly volume in April 2026; the separately operated Polymarket US, launched in early 2026, cleared roughly $1.3 billion in the same window (TRM Labs, 2026). Sports and politics dominate the order flow.
​

Kalshi is a CFTC-registered Designated Contract Market that lists event contracts in dollars through a custodial broker model. Kalshi reported around $52 billion in cumulative event-contract volume as of March 2026 (Britannica, 2026), with sports markets driving roughly 87% of the trailing-12-month figure.
​

Limitless is an onchain prediction market on Base that runs hourly binary markets on BTC and ETH price moves, settled by Pyth feeds. The venue uses a central limit order book rather than an AMM, which it argues gives tighter pricing on short-duration markets where AMM impermanent-loss risk would be high.
​

Augur was the original 2015 onchain prediction market and the first dApp deployed on Ethereum. Augur v2 (2020) replaced REP with REPv2 and added stablecoin settlement. Active trading on Augur has been thin for years; stewardship moved to the Lituus Foundation in 2025, with a forking-mechanism test on the roadmap.
​

Three limitations recur across venues. Thin liquidity on long-tail markets means prices on small or obscure questions can swing wildly on small orders. Anything outside the top few categories (US elections, major sports, crypto price moves) tends to have order-book depth measured in thousands of dollars, not millions. Forecast accuracy degrades when the marginal trader can move the price without being matched by informed counterparties.
​

Oracle disputes are the second failure mode. Optimistic-oracle resolution depends on someone showing up to dispute an incorrect assertion. Polymarket has seen multiple contested resolutions where the "common knowledge" answer differed from the strict reading of the resolution criteria. The UMA DVM exists to handle this, but it pushes resolution time from hours to days and forces tokenholders to adjudicate questions outside their expertise.
​

Regulatory treatment varies by jurisdiction, and the variance is genuinely a moving target. The CFTC's posture on sports event contracts shifted in early 2025; ongoing Congressional Research Service briefings describe active policy debate, and individual states have taken differing positions. This article does not opine on what is permitted where. Users should read each venue's published terms and their own jurisdiction's rules before participating.
​

Prediction markets that settle in stablecoins create the same cross-chain UX problem every other onchain venue creates: the user needs USDC on the specific chain the venue runs on. Polymarket needs USDC on Polygon. Limitless needs USDC on Base. Stablecoin routing infrastructure abstracts that step so a trader holding USDC on any supported chain can fund a position without a manual bridge. The same routing layer powers tokenized outcome shares moving between venues for arbitrage, settlement, or composability.
​

Sources and methodology. Volume figures pulled from The Block and TRM Labs for April 2026. Mechanism descriptions verified against Polymarket documentation and academic literature. Figures refresh quarterly; specific volumes change daily.
​