Skip to main content

What are Prediction Markets? Mechanisms, Impact, and Opportunities

· 10 min read
Dora Noda
Dora Noda
Software Engineer

Prediction markets (the term favored in research and enterprise contexts) and betting markets (the more common consumer framing) are two sides of the same coin. Both allow participants to trade contracts whose final value is determined by the outcome of a future event. In the U.S. regulatory framework, these are broadly referred to as event contracts—financial derivatives with a payoff tied to a specific, observable event or value, such as an inflation report, a storm's intensity, or an election result.

The most common format is the binary contract. In this structure, a "Yes" share will settle to $1 if the event happens and $0 if it does not. The market price of this "Yes" share can be interpreted as the collective's estimated probability of the event occurring. For example, if a "Yes" share is trading at $0.63, the market is signaling an approximate 63% chance that the event will happen.

Types of Contracts​

  • Binary: A simple Yes/No question about a single outcome. Example: “Will the BLS report Core CPI YoY be ≥ 3.0% for December 2025?”
  • Categorical: A market with multiple, mutually exclusive outcomes where only one can be the winner. Example: “Who will win the election for Mayor of New York City?” with options for each candidate.
  • Scalar: A market where the outcome is on a continuous spectrum, often with payouts bucketed into ranges or determined by a linear formula. Example: “How many interest rate cuts will the Federal Reserve announce in 2026?”

Reading Prices​

If a binary contract's "Yes" share, which pays out $1, is trading at price pp, then the implied probability is approximately pp, and the odds are p/(1−p)p / (1-p). In a categorical market with multiple outcomes, the prices of all shares should sum to approximately $1 (deviations are usually due to trading fees or liquidity spreads).

Why do these markets matter?​

Beyond simple speculation, well-designed prediction markets serve valuable functions:

  • Information Aggregation: Markets can synthesize vast amounts of dispersed knowledge into a single, real-time price signal. Studies have shown they often outperform simple benchmarks, and sometimes even traditional polls, when the questions are well-specified and the market has adequate liquidity.
  • Operational Value: Corporations have successfully used internal prediction markets to forecast product launch dates, project demand, and assess the risk of meeting quarterly objectives (OKRs). The academic literature highlights both their strengths and potential for behavioral biases, like optimism in "house" markets.
  • Public Forecasting: Long-running academic and policy programs, such as the Iowa Electronic Markets (IEM) and the non-market forecasting platform Good Judgment, demonstrate that careful question design and proper incentives can produce highly useful data for decision support.

Market Design: Three Core Mechanics​

The engine of a prediction market can be built in several ways, each with distinct characteristics.

1) Central Limit Order Books (CLOB)​

  • How it works: This is the classic exchange model where traders post "limit" orders to buy or sell at specific prices. An engine matches buy and sell orders, creating a market price and visible order depth. Early on-chain systems like Augur utilized order books.
  • Pros: Familiar price discovery for experienced traders.
  • Cons: Can suffer from thin liquidity without dedicated market makers to constantly provide bids and asks.

2) LMSR (Logarithmic Market Scoring Rule)​

  • Idea: Developed by economist Robin Hanson, the LMSR is a cost function-based automated market maker that always quotes prices for all outcomes. A parameter, bb, controls the market's depth or liquidity. Prices are derived from the gradient of the cost function: C(mathbfq)=blnsum_ieq_i/bC(\\mathbf{q})=b\\ln\\sum\_i e^{q\_i/b}.
  • Why it’s used: It offers elegant mathematical properties, bounded loss for the market maker, and gracefully supports markets with many outcomes.
  • Cons: Can be computationally intensive and therefore gas-heavy to implement directly on-chain.

3) FPMM/CPMM (Fixed/Constant Product AMM)​

  • Idea: This model adapts the popular constant product formula (xtimesy=kx \\times y = k) from DEXs like Uniswap to prediction markets. A pool is created with tokens representing each outcome (e.g., YES tokens and NO tokens), and the AMM provides continuous price quotes.
  • Where used: Gnosis's Omen platform pioneered the use of the FPMM for conditional tokens. It is practical, relatively gas-efficient, and simple for developers to integrate.

Examples and the Current U.S. Landscape (August 2025 Snapshot)​

  • Kalshi (U.S. DCM): A federally regulated exchange (Designated Contract Market) that lists a variety of event contracts. After favorable district and appellate court rulings in 2024 and the CFTC's subsequent decision to drop its appeal in 2025, Kalshi has been able to list certain political and other event contracts, though the space remains subject to ongoing policy debates and some state-level challenges.
  • QCX LLC d/b/a Polymarket US (U.S. DCM): On July 9, 2025, the CFTC designated QCX LLC as a Designated Contract Market. Filings indicate the company will operate under the assumed name "Polymarket US." This creates a regulated pathway for U.S. users to access event contracts, complementing Polymarket's global on-chain platform.
  • Polymarket (Global, On-chain): A leading decentralized platform that uses the Gnosis Conditional Token Framework (CTF) to create binary outcome tokens (ERC-1155). Historically, it blocked U.S. users following a 2022 settlement with the CFTC, but it is now moving toward a regulated U.S. presence via QCX.
  • Omen (Gnosis/CTF): A fully on-chain prediction market platform built on the Gnosis stack, using an FPMM mechanism with conditional tokens. It relies on community governance and decentralized arbitration services like Kleros for resolution.
  • Iowa Electronic Markets (IEM): A long-running, university-operated market for academic research and teaching, using small stakes. It serves as a valuable academic baseline for market accuracy.
  • Manifold: A popular "play-money" social prediction market site. It is an excellent environment for experimenting with question design, observing user experience patterns, and fostering community engagement without financial risk.

Note on Regulation: The landscape is evolving. In May 2024, the CFTC issued a proposed rule that sought to categorically prohibit certain event contracts (related to elections, sports, and awards) from being listed on CFTC-registered venues. This proposal sparked an active debate that overlapped with the Kalshi litigation and subsequent agency actions. Builders and users should always check the current rules.

Under the Hood: From Question to Settlement​

Building a prediction market involves several key steps:

  1. Question Design: The foundation of any good market is a well-phrased question. It must be a clear, testable prompt with an unambiguous resolution date, time, and data source. For example: “Will the Bureau of Labor Statistics report Core CPI ≥ 3.0% YoY for December 2025 in its first official release?” Avoid compound questions and subjective outcomes.
  2. Resolution: How will the truth be determined?
  • Centralized Resolver: The platform operator declares the outcome based on the pre-specified source. This is fast but requires trust.
  • On-chain Oracle/Dispute: The outcome is determined by a decentralized oracle, with a dispute process (like community arbitration or token-holder voting games) as a backstop. This offers credible neutrality.
  1. Mechanism: Which engine will power the market?
  • Order book: Best if you have dedicated market-making partners who can ensure tight spreads.
  • AMM (FPMM/CPMM): Ideal for "always-on" liquidity and simpler on-chain integration.
  • LMSR: A strong choice for multi-outcome markets, but requires managing gas/compute costs (often via off-chain computation or an L2).
  1. Collateral & Tokens: On-chain designs often use the Gnosis Conditional Token Framework, which tokenizes each potential outcome (e.g., YES and NO) as distinct ERC-1155 assets. This makes settlement, portfolio management, and composability with other DeFi protocols straightforward.

How accurate are these markets, really?​

A large body of evidence across many domains shows that market-generated forecasts are typically quite accurate and often outperform moderate benchmarks. Corporate prediction markets have also been shown to add value, though they can sometimes exhibit house-specific biases.

It's also important to note that forecasting platforms without financial markets, like Metaculus, can also produce highly accurate results when incentives and aggregation methods are well-designed. They are a useful sibling to markets, especially for long-horizon questions or topics that are difficult to resolve cleanly.

Risks and Failure Modes​

  • Resolution Risk: The market's outcome can be compromised by ambiguous question wording, unexpected data revisions from a source, or a disputed result.
  • Liquidity & Manipulation: Thinly traded markets are fragile and their prices can be easily moved by large trades.
  • Over-interpretation: Prices reflect probabilities, not certainties. Always account for trading fees, bid-ask spreads, and the depth of liquidity before drawing strong conclusions.
  • Compliance Risk: This is a heavily regulated space. In the U.S., only CFTC-regulated venues may legally offer event contracts to U.S. persons. Platforms operating without proper registration have faced enforcement actions. Always check local laws.

For Builders: A Practical Checklist​

  1. Start with the Question: It must be a single, falsifiable claim. Specify who, what, when, and the exact resolution source.
  2. Choose a Mechanism: Order book (if you have makers), FPMM/CPMM (for set-and-forget liquidity), or LMSR (for multi-outcome clarity, minding compute costs).
  3. Define Resolution: Will it be a fast centralized resolver or a credibly neutral on-chain oracle with a dispute process?
  4. Bootstrap Liquidity: Seed the market with initial depth. Consider offering incentives, fee rebates, or working with targeted market makers.
  5. Instrument the UX: Clearly display the implied probability. Expose the bid/ask spread and liquidity depth, and warn users about low-liquidity markets.
  6. Plan Governance: Define an appeals window, require dispute bonds, and establish emergency procedures for handling bad data or unforeseen events.
  7. Integrate Cleanly: For on-chain builds, the Gnosis Conditional Tokens + FPMM combination is a proven path. For off-chain applications, use a regulated venue’s API where permitted.
  8. Mind Compliance: Keep a close watch on the CFTC’s evolving rulemaking on event contracts and any relevant state-level regulations.

Glossary​

  • Event Contract (U.S. term): A derivative whose payoff is contingent on the outcome of a specified event; often binary (Yes/No).
  • LMSR: Logarithmic Market Scoring Rule, a type of AMM known for its bounded loss properties.
  • FPMM/CPMM: Fixed/Constant Product Market Maker, an AMM model adapted from DEXs for trading outcome tokens.
  • Conditional Tokens (CTF): A Gnosis-developed framework for issuing ERC-1155 tokens that represent positions in an outcome, enabling composable settlement.

Responsible Use & Disclaimer​

Nothing in this article constitutes legal, tax, or investment advice. In many jurisdictions, event contracts are closely regulated and may be treated as a form of gaming. In the U.S., it is critical to review CFTC rules and any state-level positions and to use registered venues where required.

Further Reading (Selected)​

Share on Twitter