Prediction Market Arbitrage Bot
Cross-platform arbitrage bot for Polymarket, Kalshi, and Predict.fun
Problem
Prediction markets fragment across Polymarket (Polygon), Kalshi (centralized, CFTC-regulated), and Predict.fun (BSC). Equivalent markets on the same real-world event routinely price differently, and within a single platform the YES+NO sum on binary markets can drift below 1.00. Capturing those inefficiencies manually is infeasible — the spreads close in seconds and require simultaneous order placement on independent venues.
Approach
- Four complementary strategies rather than a single arbitrage path: internal (YES+NO<1.00 on one venue), cross-platform (equivalent markets on different venues), dutching (equal-profit betting across correlated outcomes), and rebate farming (maker-order rebate capture).
- Bun runtime for low-latency TypeScript execution across the parallel pricing feeds.
- viem v2 for on-chain order signing on Polygon (Polymarket CLOB) and BSC (Predict.fun); direct API integration for Kalshi’s centralized order book.
- Real-time spike detector as a separate signal source for sudden repricing events that briefly open arbitrage windows.
- Telegram alerts + CSV/dashboard export so the operator can audit executions and tune thresholds.
Implementation
Arbitrage Strategies
- Internal Arbitrage: detects YES+NO price sum < 1.00 within a single platform.
- Cross-Platform Arbitrage: identifies price discrepancies for equivalent markets across Polymarket, Kalshi, and Predict.fun.
- Dutching: multi-outcome equal-profit betting across correlated markets.
- Rebate Farming: captures maker-order rebates through strategic limit order placement.
Monitoring & Alerts
Real-time spike detection bot for sudden price movements. Telegram alert integration for opportunity notifications. CSV export and dashboard for performance tracking.
Execution
Bun runtime for high-performance TypeScript execution. viem v2 for on-chain transaction handling on Polygon and BSC. Order management across centralized (Kalshi) and decentralized (Polymarket, Predict.fun) venues through a unified interface.
Outcome
- Automated arbitrage detection and execution across the three major prediction-market platforms.
- Four distinct strategy modes covering intra-venue, cross-venue, dutching, and rebate capture.
- Telegram-push opportunity alerts with CSV/dashboard performance tracking.
Technologies
- Runtime: Bun, TypeScript
- On-chain: Viem v2, Polygon, BSC
- Platforms: Polymarket CLOB, Kalshi API, Predict.fun
- Monitoring: Telegram Bot API, CSV dashboards