Cross-Market Prediction Arbitrage Scanner

The strategy continuously scans binary prediction markets across multiple platforms (Polymarket, Kalshi, PredictIt) for the same underlying event. When the sum of probabilities on complementary outcomes is less than 1 (violating no-arbitrage), the bot simultaneously buys the underpriced outcome on one platform and sells the overpriced outcome on another, locking in a risk-free profit. Entry is automatic when the discrepancy exceeds a configurable threshold (e.g., 2%). The bot uses fractional Kelly sizing to manage risk and requires a backtested out-of-sample edge before executing live trades. Implementation is in Python asyncio with a production-grade scanner and paper-trading mode.