Prediction Market Smart Contract (Polymarket-Style)

You launch a DeFi protocol where users bet on event outcomes — from sports matches to ETH price. The first version of your contract in JavaScript with Web3 fails under load: reentrancy in redeem, oracle manipulation leads to loss of funds, and gas cost on the AMM exceeds the 15 million block limit. Sound familiar? We rewrite such contracts from scratch.

How Prediction Markets Work on Blockchain

Each market is a pair of conditional tokens: YES and NO. A user deposits 1 USDC and receives 1 YES + 1 NO token via the Conditional Tokens Framework (CTF) from Gnosis. Then they sell the unwanted outcome on an AMM or CLOB. After resolution, the winning token is redeemed for 1 USDC, the losing token for 0.

Conditional Tokens Framework (CTF) — an ERC-1155 standard for conditional tokens. Each outcome is represented as a position with a unique positionId. At resolution, CTF allows redeeming winning positions.

// Gnosis CTF interface interface IConditionalTokens { function prepareCondition( address oracle, bytes32 questionId, uint outcomeSlotCount ) external; function reportPayouts( bytes32 questionId, uint[] calldata payouts ) external; function redeemPositions( IERC20 collateralToken, bytes32 parentCollectionId, bytes32 conditionId, uint[] calldata indexSets ) external; } 

AMM vs CLOB: What to Choose for Your Market

Polymarket uses CLOB for speed. For a fully decentralized market, AMM is better. LMSR is mathematically elegant, but Constant Product AMM is about 3x cheaper in gas — a good choice for markets with balanced liquidity.

Why AMM is Better for a Fully Decentralized Market

LMSR (Logarithmic Market Scoring Rule) — the classic AMM for prediction markets. The price depends on the quantity of YES and NO tokens sold:

price_yes = e^(q_yes/b) / (e^(q_yes/b) + e^(q_no/b)) 

where b is the liquidity parameter. LMSR guarantees that the market maker always accepts bets, and the maximum loss is bounded by b * ln(2). Constant Product AMM (like Uniswap v2) is simpler in gas, but prices are less accurate near 0% or 100%. We use a modified constant product with a price range limit [0.02, 0.98] — this protects against infinite losses at extreme outcomes.

Oracle and Resolution: Where the Main Complexity Hides

A prediction market without a reliable oracle is a market with manual arbitrage, always under threat. Three approaches:

• Chainlink Data Feeds — for financial markets (ETH price above $5000?). Deterministic, decentralized, but covers only finance.
• UMA Optimistic Oracle — for subjective questions (election outcome). Proposal + dispute period + dispute through UMA token holders. 2–48 hours delay, but works for any question.
• Custom multisig oracle — a set of trusted parties (5/9 multisig) votes on the outcome. Centralized but transparent and fast. For private markets.

contract PredictionMarket { struct Market { bytes32 conditionId; address oracle; uint256 endTime; uint256 resolutionTime; MarketStatus status; uint128 yesReserve; uint128 noReserve; uint256 totalVolume; } enum MarketStatus { Open, Closed, Resolved, Disputed } mapping(bytes32 => Market) public markets; // AMM pricing function getPrice(bytes32 marketId, bool isYes) public view returns (uint256) { Market storage m = markets[marketId]; uint256 yesR = m.yesReserve; uint256 noR = m.noReserve; // constant product: price_yes = noR / (yesR + noR) in fixed point return (noR * 1e18) / (yesR + noR); } } 

How to Protect Against Manipulation and Attacks

• Oracle manipulation. If a market resolves based on an on-chain price at a specific moment, a flash loan attack is possible. Protection: TWAP over 24–48 hours instead of spot price, multiple independent sources.
• Front-running of resolution. Someone learns the outcome before official resolution and buys tokens at old prices. Solution: stop trading X hours before resolution.
• Griefing through spam disputes. In optimistic oracle systems, an attacker disputes every resolution. Protection: bond requirement for disputes (collateral lost if the dispute fails).
• Reentrancy during redeem. CTF.redeemPositions transfers tokens before updating state. Use Check-Effects-Interactions + ReentrancyGuard.
• Wrong conditionId. CTF uses keccak256(oracle, questionId, outcomeSlotCount). Verify conditionId twice before deployment.

Governance and Market Creation

Who can create markets? Options:

• Permissioned. Only whitelisted operators. Centralized but protects against spam markets.
• Permissionless with a bond. Anyone who pays a bond. Bond is returned on correct resolution.
• DAO governance. Voting for each market. Slow but decentralized.

For MVP — permissioned with a roadmap to DAO.

Development Stack

• Foundry — primary tool. Fuzz tests for AMM math.
• Gnosis CTF — use the ready-made implementation.
• Chainlink — oracle for financial markets.
• OpenZeppelin — AccessControl, ReentrancyGuard, Pausable.
• Slither + Echidna — static analysis and property-based testing.