Cross-Platform Polymarket/Kalshi Arbitrage with Kelly Sizing

How It Works Polymarket and Kalshi often list markets on the same underlying event (elections, economic data, sports). When the implied probability of an outcome differs materially between the two platforms, a trader can buy YES on the cheaper platform and YES on the NO side of the same outcome on the more expensive platform (or equivalent), locking in a spread regardless of resolution.

Entry/Exit Conditions An async Python scanner continuously polls both platforms' APIs for matching markets and computes the implied probability gap. A trade is entered when the spread exceeds a configurable threshold sufficient to cover transaction costs. Position sizing uses the Kelly criterion applied to the net edge, capped by risk management parameters to avoid over-sizing on uncertain probability estimates.

Verified Results The repo description mentions an embedded analytics dashboard, suggesting live or paper-trading performance tracking is built in. No specific PnL results are published in the available description.

Implementation Notes The system is built in async Python, enabling concurrent polling of both APIs to minimize the latency gap between price observation and execution. Risk-managed execution suggests stop-loss or maximum drawdown constraints are implemented alongside Kelly sizing. The analytics dashboard provides ongoing monitoring of edge decay and fill quality.

Risks Market definitions may not perfectly match across platforms (different resolution sources, timing), creating basis risk. Regulatory differences between Polymarket (crypto) and Kalshi (CFTC-regulated) can cause one platform to delist or resolve differently. API rate limits on either platform can delay detection of opportunities. Kelly sizing amplifies losses if edge estimates are miscalibrated.