VolHarvest: Autonomous LLM Options Trader

Traditional algorithmic trading bots make simple directional bets on whether a stock will go up or down. We wanted to build something far more advanced. VolHarvest AI is a fully autonomous options trading system designed to profit not from direction, but from pure market volatility. The system operates via a continuous autonomous loop on a watchlist of high-beta tech stocks (like NVDA and TSLA). First, it uses yfinance to scrape the latest real-time macro news and headlines. Second, it passes these headlines to an open-source LLM via NVIDIA's NIM API (specifically the massive 550B Nemotron model), prompting it as a Quantitative Analyst to predict the magnitude of the market reaction: EXTREME or SIDEWAYS. If the LLM predicts an extreme reaction (e.g., due to an earnings report or acquisition), VolHarvest uses the Alpaca CLI to fetch the live Options Chain. It mathematically parses the JSON data to find the cheapest valid Options Straddle (buying a Call and a Put at the exact same strike price). Finally, it autonomously submits the multi-leg limit orders directly to the Alpaca Paper Trading environment. We also built a beautiful Streamlit Web Dashboard that tracks the agent's live reasoning and active portfolio. VolHarvest strictly fulfills every hackathon requirement: it is an autonomous agent, heavily leverages the Alpaca CLI for execution, trades complex multi-leg options, and integrates partner technology via NVIDIA's OpenAI-compatible endpoints.