Prediction Markets

02 September 2026

Prediction markets let people trade shares in the outcome of future events - elections, sports championships, crypto protocol upgrades, even whether a specific temperature record will be broken. On-chain prediction markets encode those shares as digital tokens, resolve them against real-world data, and settle trades without a central bookie. This page surveys the entire system: how markets are created, how positions are priced and traded, how outcomes get settled, what can go wrong, and which tools exist for each step. The spoke pages beneath this one go deep on each mechanism; you will find handoffs to them throughout.

Market creation and tokenization

Every prediction market starts with a question. Someone - a market creator - proposes an unambiguous binary question such as "Will the Federal Reserve cut interest rates at the May 2025 FOMC meeting?" or a multi-outcome question like "Which party will win the 2025 German federal election?" The creator deposits collateral, typically USDC or DAI, and the platform mints two classes of outcome tokens: one for "Yes" and one for "No" in a binary market, or one per candidate in a multi-outcome market. These tokens are not ordinary ERC-20s. Most modern platforms, including Polymarket and the Omen interface, use the Gnosis conditional token framework (ERC-1155 outcome shares) to bundle outcomes under a single contract. The spoke page How the Gnosis conditional token framework powers prediction market shares explains why that matters for composability and gas efficiency.

To deter spam and bad-faith questions, most platforms require a market creation bond or validity deposit. On Augur, that bond can be hundreds of DAI; on Polymarket, the cost varies by market type and is subject to change. The bond is returned after the market resolves correctly, but it can be forfeited if the creator lists an unanswerable or duplicate question. The spoke How market creation bonds and validity deposits work details the amounts and forfeiture conditions.

Trading mechanisms: amms and order books

Once a market exists, traders need a way to buy and sell outcome shares. On-chain prediction markets use two fundamentally different trading models, and sometimes both on the same platform.

Automated market makers (AMMs) are the simpler model for binary outcomes. A liquidity pool holds both outcome tokens and a collateral reserve. When you buy "Yes" shares, the AMM's pricing curve - most commonly a constant product formula or a logarithmic market scoring rule - adjusts the price upward as "Yes" demand increases. The resulting share price maps directly to a probability: if "Yes" costs $0.65, the market implies a 65% chance of that outcome. The spoke How automated market makers work for binary prediction markets walks through the math and the real-world liquidity constraints.

Order book models let traders place limit and market orders against each other, just like on a centralized exchange. SX Network and certain Polymarket sub-markets use this approach. Order books offer tighter spreads when liquidity is deep, but they require continuous market-making and suffer from thin order books on niche events. The comparison spoke Order book versus AMM trading in on-chain prediction markets spells out the practical trade-offs.

Both models rely on real-time odds calculation from share price. Platforms display percentages like "Yes 72%," but those numbers are derived from the last traded price or the AMM curve's mid-price. The odds are not forecasts; they are the market's current consensus price. The spoke How real-time odds are calculated from prediction market share prices explains the conversion from on-chain data to the percentage you see in a browser.

Resolution oracles and settlement

A prediction market is worthless if you cannot cash out winning shares. The process of determining the actual outcome and triggering payouts is called resolution, and it is the most fragile part of the entire system.

Resolution oracles submitting settlement data on-chain are the only path from a real-world event to a settled smart contract. An oracle - human or automated - watches the event, confirms the result, and submits a data packet to the market contract. That packet might say "outcome index 0" for "Yes" or include a scalar value like "2.7" for "How many inches of rain will fall in Chicago on July 4?" The spoke How resolution oracles submit settlement data to prediction markets covers the actual submission flow and the tools involved, including the UMA Optimistic Oracle and Chainlink data feeds for sports event resolution.

Once the oracle submits data, a challenge period begins. During that window - typically 24 hours to a week - anyone can dispute the resolution by posting a bond. If no dispute occurs, funds become withdrawable. If a dispute is filed, the system escalates to a decentralized court or an optimistic oracle appeal flow. The spoke What happens when a prediction market resolution is disputed explains the full escalation chain, including the UMA optimistic oracle dispute process and the Kleros decentralized court appeal flow. The spoke What happens when a prediction market oracle resolves incorrectly covers the worst-case scenario: the oracle was wrong, the dispute failed, and traders lose funds.

After the challenge period ends with no dispute, or after a dispute is resolved in favor of the oracle's submission, winners can redeem their outcome tokens for collateral. This is not automatic. You must call a redeemPositions function on the market contract, which burns your winning shares and returns proportional collateral. The spoke How outcome tokens are minted and redeemed in prediction markets walks through the exact transaction flow, including the common error "Outcome token redemption window closed."

Platforms, tools, and custody

Most retail prediction market activity today happens on Polymarket, which runs on Polygon and uses USDC as its settlement currency. Polymarket charges zero trading fees (as of this writing) and handles custody through a hybrid model: the market contract holds collateral, but your wallet retains control of your outcome tokens. The spoke How trading on Polymarket works from wallet connection to settlement is the step-by-step tutorial for that specific platform.

Other platforms serve different niches. Augur Turbo is a lower-friction version of the original Augur protocol, but it still requires REP staking for dispute resolution. Azuro Protocol focuses on sports markets with AMM liquidity. SX Network (formerly SportX) uses an order book model. The comparison spoke Polymarket vs Augur prediction market platforms compared breaks down the fee structures, liquidity depth, and resolution guarantees.

Collateral custody is a recurring concern. In non-custodial models, your collateral stays in your wallet until you approve a transfer to a market contract. In custodial models, the platform holds a pooled balance. The spoke How collateral custody works in on-chain prediction markets contrasts the three main custody models - self-custody, platform custody, and hybrid - with named examples for each.

Risks in Practice

The inventory includes fifteen distinct risks. The most severe are oracle manipulation or incorrect resolution data and smart contract exploit draining market collateral. A malicious oracle can submit a false result and, if the dispute system fails, drain all winning payouts. A contract bug can let an attacker drain the entire liquidity pool. The spoke Smart contract risks in on-chain prediction markets explained lists the major exploit categories and tells you which questions to ask before depositing into any market.

Low liquidity causing inability to exit position is the most common day-to-day frustration. You place a trade, the market moves against you, and when you try to sell, the AMM shows "Insufficient liquidity to fill order at this price." The spoke How to fix insufficient liquidity errors in prediction market trading explains the slippage tolerance setting, the "Insufficient liquidity" and "Slippage tolerance exceeded" errors, and how to split large orders or switch to an order book market.

Stablecoin depeg of USDC collateral is a structural risk that affects every platform using USDC. If USDC loses its dollar peg, your collateral and payouts lose value simultaneously. The spoke USDC vs DAI as collateral for prediction market trading compares the two stablecoins' risk profiles, including depeg history and recovery mechanisms.

Misconceptions Worth Clearing

Three misconceptions appear repeatedly. First: "Prediction markets are gambling and illegal everywhere." In many jurisdictions, including the United States, on-chain prediction markets operate in a legal gray area except where explicitly banned (e.g., certain U.S. states limit access to Polymarket). They are not gambling in the traditional sense because the outcome is not random - it is determined by a real event - but regulators often treat them as derivative contracts. The legal landscape is volatile and platform-specific.

Second: "Share price equals percentage probability exactly." It does not. The share price is a market-clearing price that reflects the last marginal buy or sell. It includes the liquidity provider's spread, potential slippage, and any market inefficiency. A 72-cent "Yes" share implies the market is currently pricing a 72% probability, but that number is not a prediction; it is a transaction price.

Third: "Once a market resolves, funds are instantly withdrawable." The challenge period blocks withdrawals. Most platforms show "Market resolved" but still prevent redemptions until the dispute window closes. The error "Withdrawal locked during dispute period" is a normal, expected state.

The Whole Picture and Where to Go Next

Prediction markets are not a single product. They are a stack of mechanisms - market creation, token minting, AMM or order book trading, oracle submission, dispute escalation, and redemption - that must all work correctly for a trade to succeed. The spoke pages below address each layer in detail.

• Start with How automated market makers work for binary prediction markets if you want to understand why share prices move the way they do.
• Read How resolution oracles submit settlement data to prediction markets if you are worried about whether a market will pay out correctly.
• Read What happens when a prediction market resolution is disputed if you want to know how the system handles mistakes.
• Read How trading on Polymarket works from wallet connection to settlement if you just want to place a trade today.
• Read Smart contract risks in on-chain prediction markets explained before depositing any significant amount.

Each of those spokes is designed to be self-contained. You can jump to the one that matches your question right now.