BioTrader AI Autonomous Trading Agent

## Inspiration Biotechnology equities (CRSP, VRTX, MRNA) are highly volatile and move dramatically based on clinical trial endpoints, statistical p-values, target protein binding affinity (Ki), and toxicity reports. Traditional quantitative trading algorithms analyze only price-volume charts and are blind to scientific literature. BioTrader AI bridges computational biology and quantitative finance by turning scientific literature parsing into an alternative data source for trading agents. ## How We Built It BioTrader AI follows a modular Model Context Protocol (MCP) architecture: 1. **Bioinformatics MCP Tools**: Continuously monitors a target biotech watchlist, extracting statistical p-values, clinical trial primary endpoint status, and target protein binding metrics (Ki in nM). 2. **Autonomous Decision Engine**: Evaluates scientific sentiment. If a clinical trial achieves statistical significance (p < 0.01) with potent binding, the agent triggers a BUY order. If FDA safety holds or high adverse event toxicity (>15%) is detected, it executes a SELL order to protect equity. 3. **Alpaca Trading API SDK**: Connects to the Alpaca Paper Trading SDK to execute real-time buy/sell orders, track equity, cash balance, and margin buying power. 4. **Trade Rationale Generator**: Powered by Qwen-2.5-72B-Instruct, the agent generates a bio-quantitative rationale explaining the biological mechanism behind every trade. ## Impact & Efficiency BioTrader AI runs on 100% CPU hardware with zero GPU requirements, making automated quantitative bio-trading accessible.