Kalshi–Polymarket Cross-Platform Arbitrage

How It Works Kalshi and Polymarket both list binary prediction markets on overlapping topics (politics, economics, crypto prices, etc.). Because the two platforms have separate liquidity pools and user bases, identical or near-identical contracts occasionally diverge in implied probability. The strategy scans both order books continuously, identifies when the sum of the best ask on one platform and the best bid on the other creates a locked profit, then executes paired trades simultaneously.

Entry / Exit Conditions Entry is triggered when the combined cost of buying YES on platform A and YES on platform B (which together guarantee a $1 payout regardless of outcome) is less than $1, net of fees. Exit is automatic at settlement: one leg pays $1 and the other pays $0, locking in the spread captured at entry. No discretionary exit decision is required.

Verified Results No on-chain or backtested P&L figures are published in any of the referenced repos. The samuel483 repo has 5 stars and 2 forks, suggesting some community validation, but no audited results are available.

Implementation Notes Multiple repos implement this in Python (lufegaga, samuel483) and TypeScript (Juanp2389). Key engineering challenges include API rate limits on both platforms, latency in order routing, and ensuring both legs fill before the price moves. The Juanp2389 bot focuses on BTC 15-minute markets, narrowing the scope to high-frequency short-duration contracts where price gaps may be more frequent.

Risks - Leg risk: one side fills and the other does not, leaving a naked directional position. - Fee erosion: USDC gas fees, platform fees, and slippage can eliminate the theoretical spread. - Contract non-equivalence: resolution rules may differ subtly between platforms, creating basis risk. - Regulatory asymmetry: Kalshi is CFTC-regulated; differing settlement interpretations are possible.