What Are Prediction Markets?

A prediction market is a place where people buy and sell shares tied to the outcome of a future event. The price of a share reflects how likely participants collectively believe that outcome is. If a share for “Candidate X wins the election” trades at $0.65, the market is saying there is roughly a 65% chance that happens — at least according to the aggregate judgment of everyone trading at that moment.

That simple mechanic has attracted economists, traders, governments, and now blockchain developers for decades. It also raises a question that still does not have a clean answer: is this sophisticated forecasting, or is it gambling with extra steps? This article covers the basics; later articles in this series dig into specific platforms and the harder questions.

How the Price–Probability Connection Works

Every share in a prediction market pays out $1 (or its equivalent) if the underlying event resolves in its favor, and $0 if it does not. That binary payoff structure means a share priced at $0.40 implies a 40% probability in the eyes of whoever is willing to buy it at that price.

This is not just a metaphor. Under certain assumptions — rational participants, competitive markets, real money at stake — prices in these markets tend to track true probabilities better than polls, expert panels, or other common forecasting methods. The people who hold wrong beliefs lose money; the people who hold correct beliefs profit. Over time this feedback loop pushes prices toward accuracy.

In practice, the assumptions are never perfectly met. Markets thin out on obscure events, prices can be manipulated by large participants, and humans are systematically irrational about low-probability outcomes. But the core logic is sound enough that several governments and large research institutions have run prediction markets internally to improve forecasts.

A Brief History

Iowa Electronic Markets (1988–present). The University of Iowa launched what is often cited as the first modern, real-money prediction market. Originally designed to forecast US election outcomes, the IEM required academic approval, kept stakes low, and was explicitly framed as a research tool. Its election forecasts routinely outperformed major polling averages. It still runs today.

Hollywood Stock Exchange (1996). A play-money market for film box-office predictions that became surprisingly accurate, demonstrating that even without real financial stakes, aggregated judgment can work well.

TradeSports / Intrade (2000s). These Dublin-based exchanges let users trade contracts on political and sporting events. Intrade became famous for its election markets and attracted serious bettors and media attention. It shut down in 2013 after regulatory pressure from US authorities and an internal financial scandal.

DARPA Policy Analysis Market (2003). A US government proposal to create a market for forecasting geopolitical events — including potential terrorist attacks — was cancelled after politicians called it morally offensive. The episode illustrated how powerful the concept is, and how controversial.

Augur (2018). The first major on-chain prediction market, built on Ethereum. Augur used a decentralized oracle system where token holders resolved disputes. It pioneered the use of smart contracts for prediction markets but struggled with low liquidity and a complicated user experience.

Polymarket (2020–present). Built on Polygon and using USDC, Polymarket became the highest-volume on-chain prediction market, especially during the 2024 US election cycle. It is non-custodial and officially restricts US users, though enforcement relies on IP blocking.

Kalshi (2021–present). The first US-regulated event-contract exchange, designated by the Commodity Futures Trading Commission (CFTC). Kalshi operates with full KYC, custodial accounts, and is legally available to US residents.

What Gets Traded

Modern prediction markets cover an enormous range of events:

• Politics: election outcomes, policy decisions, legislative votes
• Economics: interest rate decisions, GDP figures, inflation readings
• Science and technology: AI benchmark results, space mission outcomes
• Sports: match winners, championship outcomes (though sports betting regulations complicate this)
• Geopolitics: treaty signings, conflict developments
• Culture: awards show results, entertainment releases

The breadth matters because it affects how the market works. A liquid market on a major election has many informed participants and tight spreads. A thin market on a niche event may have only a handful of traders, making prices noisy and manipulation easier.

How Markets Are Structured

Most prediction markets use one of three structures:

Order books. Buyers and sellers post limit orders at prices they are willing to trade. When a buyer’s bid meets a seller’s ask, a trade executes. This is the same structure used by stock exchanges. Polymarket uses this model.

Automated market makers (AMMs). A liquidity pool holds both sides of the market, and an algorithm sets prices based on the ratio of shares in the pool. Traders always have a counterparty but pay a spread that grows with trade size.

Logarithmic Market Scoring Rules (LMSR). A mathematical market-maker system where a central entity always quotes prices and absorbs all trades. Designed for thin markets where natural buyers and sellers may not show up at the same time. Used in several academic and corporate internal prediction markets.

Resolution: How Markets Settle

Every prediction market must have a resolution mechanism — a way to determine which outcome actually occurred and distribute funds accordingly.

Centralized platforms like Kalshi use their own staff and defined resolution rules to settle markets. On-chain platforms like Polymarket rely on oracle systems — external data providers that report real-world outcomes to the blockchain. Polymarket uses UMA Protocol’s optimistic oracle, which allows anyone to dispute a proposed resolution during a challenge window.

Resolution is where many problems arise. Ambiguously worded questions, unusual outcomes, and oracle failures have all caused disputes that cost traders money. See our article on prediction market risks and manipulation for detail on how this plays out.

The Risk Picture

Prediction markets are financially real. You can lose everything you stake. Unlike a poker table where the worst case is losing your buy-in gradually, a prediction market share can go to $0 the moment a result is confirmed — which can happen instantaneously. Key risks include:

• Total loss. You hold shares worth $0 if the event resolves against you.
• Liquidity risk. You may not be able to exit a position before resolution at any reasonable price.
• Resolution disputes. Ambiguous events can resolve in unexpected ways.
• Platform risk. Smart contract bugs, platform insolvency, or regulatory shutdown can freeze funds.

The “wisdom of crowds” is real, but it is not a guarantee. Markets have been wrong on major events, and being right about an outcome does not protect you from losing money if you paid too high a price for your shares.

Is It Trading or Gambling?

This question comes up constantly and deserves a serious answer — which is why there is a dedicated article on it in this series. The short version: it depends heavily on how you are participating. Informed traders with genuine research-based edges resemble traders more than gamblers. Casual participants chasing the excitement of a live election market resemble gamblers more than traders. The structure supports both behaviors simultaneously, and the regulatory treatment varies by jurisdiction and by specific platform.

Prediction markets are genuinely interesting and genuinely risky. They represent a real contribution to how societies aggregate information — and a real way for individuals to lose significant amounts of money. Both things are true.

For a deeper look at the mechanics, read how prediction market prices and odds work, or visit our responsible gambling page if you are evaluating whether to participate.