Prediction Market Trading Bot
What It Does
A Python-based automated trading system for prediction markets. The bot runs multiple strategies that pull real-time data from external sources (NOAA weather forecasts, ESPN live sports probabilities) and compare them against prediction market prices. When the bot detects a gap between what the data says and what the market is pricing, it places trades automatically. It supports both paper trading and live execution, with a full risk management layer handling position sizing, daily loss limits, and cooldowns. All state is persisted to Supabase so the system survives restarts. A separate Next.js dashboard polls the database and displays open positions, P&L, trade history, and strategy performance in real time.
Key Features
- Multiple automated strategies comparing external data sources against market prices
- Weather strategy: NOAA forecasts vs. market-priced temperature and precipitation contracts
- Sports strategies: ESPN live and pre-game win probabilities vs. market prices
- Full risk management: position sizing, daily loss limits, cooldowns, and longshot protection
- Paper trading and live execution modes
- Real-time dashboard showing open positions, P&L, trade history, and strategy performance
- All state persisted to Supabase for crash recovery and restart resilience
- Remote kill switch via dashboard
Why I Built It
Prediction markets are a data problem. If you can get better data faster than the market is pricing, there are opportunities. I wanted to build a system that finds and acts on those gaps automatically.
What I Learned
Learned how to build a stateful trading system with crash recovery, integrate multiple real-time data APIs, implement risk management controls, and design a live monitoring dashboard for an autonomous system.