Prediction Market

What Is a Prediction Market?

A prediction market is an exchange where participants trade contracts tied to the outcomes of future events. Each contract (called an outcome token) trades between $0 and $1, and its market price represents the crowd's consensus probability that a particular outcome will happen. If the event occurs, the token settles at $1. If it does not, the token settles at $0.

Traditional prediction markets have existed since the early 1990s with platforms like the Iowa Electronic Markets, which let participants trade on US presidential elections. Crypto-native prediction markets brought this concept on-chain, using smart contracts to eliminate the need for a trusted central operator. Augur launched on Ethereum in 2015 as one of the first decentralized prediction markets, and Polymarket emerged in 2020 to become the dominant platform, processing over $3.6 billion in trading volume on the 2024 US presidential election alone.

The fundamental insight behind prediction markets is that prices aggregate information more efficiently than surveys or expert panels. When real money is at stake, participants with superior knowledge are incentivized to trade on it, pushing prices toward accurate probabilities. This mechanism is often called the "wisdom of crowds."

How It Works

Prediction markets operate through a straightforward tokenization mechanism. For a binary event (yes or no), the market creates two tokens: a YES token and a NO token. Together, one YES token and one NO token always equal $1 in value, because exactly one outcome must occur.

1. A market creator defines the event, possible outcomes, and resolution criteria (for example, "Will Bitcoin exceed $150,000 by December 31, 2026?")
2. Participants deposit funds (typically a stablecoin like USDC) to mint a complete set of outcome tokens: one YES and one NO token per $1 deposited
3. Traders buy and sell these tokens on the open market based on their beliefs about the outcome's probability
4. When the event resolves, an oracle reports the outcome on-chain
5. Holders of the winning token redeem each token for $1, while the losing token becomes worthless

If a YES token trades at $0.65, the market implies a 65% probability that the event will occur. A trader who believes the true probability is 80% would buy YES tokens at $0.65, expecting to profit $0.35 per token when the event resolves. This arbitrage dynamic pushes prices toward accurate probability estimates.

Order Book vs. AMM-Based Markets

Prediction markets use two primary mechanisms for price discovery and trade matching:

Polymarket uses a central limit order book (CLOB) on Polygon with off-chain order matching and on-chain settlement, achieving tighter spreads and lower slippage than AMM alternatives. Earlier platforms like Augur and Gnosis's Omen used AMM models, sometimes based on the Logarithmic Market Scoring Rule (LMSR) designed by economist Robin Hanson specifically for prediction markets.

Oracle Resolution

Every prediction market needs a mechanism to determine which outcome actually occurred. This is the oracle problem applied to event outcomes rather than price feeds. Different platforms use different approaches:

• Optimistic oracle (Polymarket/UMA): a proposer asserts the outcome and posts a bond. If no one disputes during the challenge window, the outcome is accepted. If disputed, UMA token holders vote on the correct resolution.
• Decentralized reporter model (Augur): REP token holders stake on outcomes. Disputes trigger escalation rounds requiring progressively larger stakes, up to a full protocol fork as the ultimate backstop.
• Committee-based resolution: some platforms use designated reporters or multi-signature committees for faster, simpler resolution at the cost of decentralization.

Oracle resolution is the most critical security surface in any prediction market. Oracle manipulation can allow attackers to steal funds by resolving markets incorrectly, making the oracle design a key differentiator between platforms.

Technical Implementation

On-chain prediction markets typically implement outcome tokens as ERC-1155 or ERC-20 tokens. The core smart contract logic handles minting, trading, and redemption:

// Simplified prediction market contract logic

// Mint a complete set: deposit $1, receive 1 YES + 1 NO token
function mintCompleteSet(marketId, amount) {
  transferFrom(msg.sender, vault, amount * 1e6);  // USDC deposit
  mint(msg.sender, YES_TOKEN, amount);
  mint(msg.sender, NO_TOKEN, amount);
}

// Redeem winning tokens after resolution
function redeem(marketId, amount) {
  require(market.resolved == true);
  burn(msg.sender, market.winningToken, amount);
  transfer(msg.sender, amount * 1e6);  // USDC payout
}

// Redeem a complete set (1 YES + 1 NO = $1) before resolution
function redeemCompleteSet(marketId, amount) {
  burn(msg.sender, YES_TOKEN, amount);
  burn(msg.sender, NO_TOKEN, amount);
  transfer(msg.sender, amount * 1e6);
}

The complete-set minting and redemption mechanism is what guarantees the YES + NO invariant: because anyone can mint or redeem a pair for exactly $1, arbitrageurs enforce that the two token prices always sum to $1 on the open market.

Prediction Markets vs. Polls

Prediction markets have repeatedly demonstrated superior forecasting accuracy compared to traditional polling. The Iowa Electronic Markets, an academic prediction market at the University of Iowa, outperformed polls 74% of the time across decades of US presidential election forecasting.

The 2024 US presidential election provided a dramatic example: Polymarket consistently priced a Trump victory at higher probabilities than most polling averages suggested, with some polls showing a toss-up or a slight Harris lead. The market's signal proved more accurate.

Several structural advantages explain this outperformance:

• Real-time information aggregation: markets update continuously as new information emerges, while polls are periodic snapshots
• Skin in the game: participants risk real capital, discouraging uninformed or performative responses that plague survey data
• Arbitrage correction: mispriced markets attract informed traders who profit by pushing prices toward accuracy
• Diverse information sources: markets aggregate private information from participants across different industries, geographies, and expertise areas

Research by economists Justin Wolfers and Eric Zitzewitz found that prediction markets are effective at aggregating dispersed information, often producing probability estimates that are well-calibrated against actual outcomes.

Major Platforms

Polymarket

Founded in 2020 by Shayne Coplan, Polymarket became the dominant crypto prediction market. The platform uses USDC for trading on Polygon, combining off-chain order matching with on-chain settlement. Polymarket hit $1.8 billion in monthly trading volume in October 2024 during the US presidential election cycle and reached $500 million in open interest at its peak.

Polymarket is not available to US users following a $1.4 million settlement with the CFTC in January 2022 for operating an unregistered trading facility. It uses UMA's optimistic oracle for outcome resolution.

Kalshi

Kalshi is a US-regulated prediction market exchange registered with the CFTC as a Designated Contract Market (DCM). In 2023, the CFTC sued Kalshi to block its congressional election contracts. A federal court ruled in Kalshi's favor in September 2024, allowing political event contracts. This landmark ruling opened the door for regulated prediction markets on political events in the United States.

Augur and Gnosis

Augur launched in 2015 as a pioneering decentralized prediction market on Ethereum, using its REP token for decentralized oracle resolution. While technologically innovative, it struggled with liquidity and user adoption. Gnosis developed the conditional token framework (based on ERC-1155) used by its Omen prediction market front-end before pivoting to focus on Gnosis Chain infrastructure.

Use Cases

Prediction markets extend well beyond political elections:

• Geopolitical and economic forecasting: markets on interest rate decisions, GDP growth, geopolitical events, and regulatory outcomes
• Cryptocurrency markets: predictions on ETF approvals, halving price impacts, protocol upgrades, and token launches
• Corporate decision-making: internal prediction markets let companies aggregate employee knowledge on project timelines, product launches, and strategic decisions
• Scientific and technological outcomes: markets on clinical trial results, AI milestones, and space mission success
• Sports and entertainment: outcomes for games, awards, and competitions, functioning as a decentralized alternative to traditional sports betting

On Bitcoin, Discreet Log Contracts (DLCs) enable trustless oracle-based contracts that could power prediction market functionality without relying on a separate blockchain. DLCs use adaptor signatures to create conditional payouts based on oracle attestations, making them a natural fit for binary outcome bets directly on the Bitcoin network. For more on how Bitcoin-native contracts work, see our research on Discreet Log Contracts.

Regulatory Considerations

Prediction markets exist in a regulatory gray zone that varies significantly by jurisdiction. In the United States, the Commodity Futures Trading Commission (CFTC) has jurisdiction over event contracts and considers many prediction market offerings to be regulated derivatives.

Key regulatory developments include:

• Polymarket's 2022 CFTC settlement ($1.4 million fine) for operating without registration, leading to its exit from the US market
• The September 2024 federal court ruling in Kalshi's favor, establishing that political event contracts can be listed on regulated exchanges
• The CFTC under the Trump administration signaling a more permissive stance toward prediction markets in 2025, including soliciting public comment on a broader event contract framework

Different jurisdictions treat prediction markets as gambling, derivatives, or novel financial instruments, and no comprehensive global framework exists. Decentralized prediction markets add complexity because they operate without a central entity that regulators can target, though front-end operators and fiat on-ramps remain points of regulatory enforcement.

Risks and Considerations

Liquidity Risk

Many prediction markets suffer from thin liquidity, especially for niche or long-dated events. Low liquidity leads to wide bid-ask spreads, high slippage, and prices that may not accurately reflect true probabilities. Even Polymarket's most active markets can see significant price impact on large orders.

Oracle Risk

The entire value proposition of a prediction market depends on accurate and honest outcome resolution. If an oracle is compromised, manipulated, or simply wrong, participants holding the correct token lose their funds. Oracle manipulation is one of the most significant attack vectors in DeFi, and prediction markets inherit this risk directly.

Market Manipulation

Wealthy participants can temporarily distort prices by placing large trades, especially in illiquid markets. During the 2024 election cycle, questions arose about whether large Polymarket traders were placing bets to influence public perception rather than to express genuine probability estimates. This highlights the tension between prediction markets as information tools and as financial instruments subject to manipulation.

Regulatory Risk

Platforms operating in uncertain regulatory environments face the risk of enforcement actions, forced shutdowns, or restrictions that could strand user funds. Polymarket's US exit after its CFTC settlement demonstrates how regulatory action can abruptly change market access.

Smart Contract Risk

Decentralized prediction markets rely on smart contract code for minting, trading, and settlement. Bugs or vulnerabilities in this code can lead to loss of funds. Users should verify that platform contracts have undergone security audits before depositing significant capital.

This glossary entry is for informational purposes only and does not constitute financial or investment advice. Always do your own research before using any protocol or technology.