This project identified and systematically exploited price discrepancies for the same real-world events across two prediction market platforms — Kalshi and Polymarket. The team built a full-stack trading system including a pairs generator with 98.2% matching accuracy across ~700,000 markets, an execution engine with guard filters, and a backtesting framework built on 477 million trades from 2024–2026. Over a 13-day paper trading window (April 16–28), the system executed 446 unique positions generating approximately $6,856 in estimated profit at a 7.68% weighted average edge, with tennis, MLB, and cricket as the top-performing sectors. Key findings include that arbitrage windows frequently held for hours due to platform-specific fee structures and asynchronous order books, and that execution speed — not signal identification — is the primary bottleneck for live deployment.