How to Find Arbitrage in Prediction Markets
Key takeaway: Prediction market arbitrage occurs when the same event is priced differently on two platforms — or when YES + NO prices on a single market sum to less than $1. These risk-free (or near risk-free) opportunities are rare but real, and understanding them makes you a sharper trader.
Prediction market arbitrage is one of the most sought-after strategies among professional traders. Unlike directional bets where you need to be right about an outcome, arbitrage profits from pricing inefficiencies — regardless of what actually happens. In this guide, we break down the mechanics, tools, and pitfalls.
What is prediction market arbitrage?
Arbitrage is the practice of simultaneously buying and selling the same asset in different markets to profit from a price difference. In prediction markets, there are two main types:
- Cross-platform arbitrage: The same event trades at different prices on Polymarket and Kalshi (e.g., YES at 42 cents on Polymarket, NO at 55 cents on Kalshi — total cost 97 cents, guaranteed $1 payout)
- Intra-market arbitrage: YES + NO shares on the same market sum to less than $1.00 (e.g., YES at 48 cents + NO at 50 cents = 98 cents). Buy both, guaranteed 2-cent profit per share
Why do arbitrage opportunities exist?
Prediction markets are fragmented across platforms with different user bases. Polymarket attracts crypto-native traders while Kalshi serves US-regulated finance. Their information sets and risk preferences differ, creating pricing gaps. Additional causes include:
- Delayed information propagation between platforms
- Different fee structures affecting effective prices
- Liquidity differences — thin markets overshoot on news events
- Withdrawal and deposit friction making capital movement slow
How to spot arbitrage opportunities
Manual monitoring is impractical for serious arb traders. Here is a systematic approach:
- Map equivalent markets — create a spreadsheet linking identical questions across platforms (Polymarket, Kalshi, Betfair, Metaculus)
- Monitor price feeds — use APIs (Polymarket's CLOB API, Kalshi's REST API) to pull mid-prices every 30 seconds
- Calculate the arb spread — if Platform A YES + Platform B NO < $1.00, there is an arbitrage. Subtract fees from both sides to get net profit
- Execute simultaneously — speed matters. Use limit orders on both sides to lock in the spread before it closes
Real-world example
During the 2024 US election, "Will Biden drop out?" traded at 32 cents YES on Polymarket and 72 cents NO on a UK exchange — a combined cost of $1.04. No arbitrage there. But two hours after the first withdrawal rumours, Polymarket moved to 58 cents while the UK exchange lagged at 65 cents NO. For a brief window, the combined cost was 58 + (100 - 65) = 93 cents — a 7-cent risk-free profit per share.
Risks and limitations
Arbitrage in prediction markets is not truly "risk-free":
- Execution risk: Prices move while you place the second leg
- Settlement risk: Different platforms may resolve the same question differently
- Capital lockup: Your funds are locked until market resolution (could be months)
- Fee erosion: Platform fees, withdrawal fees, and slippage can eat your edge
- Counterparty risk: One platform could face insolvency or regulatory action
⚠️ Always account for ALL fees (trading, withdrawal, gas) before declaring an arbitrage profitable. A 3-cent arb with 4 cents in fees is a losing trade.
Tools for prediction market arbitrage
Several tools can help identify opportunities:
- PolyGram's portfolio analytics — track positions across markets with real-time P&L at polygram.ink/analytics
- Custom scripts — Python bots using Polymarket's API to scan for cross-market price discrepancies
- Community alerts — Discord and Twitter communities share arb opportunities (though they close quickly once shared)
Ready to put arbitrage theory into practice? Start trading on PolyGram →