Phoenix Alpha: Autonomous Options Trading Desk

Algorithmic trading systems typically fail in one of two ways: simple quantitative bots buy dips blindly without understanding root causes—often catching falling knives during structural decline—while ungrounded LLM agents hallucinate trade parameters and lack deterministic risk bounds. Phoenix Alpha solves this by enforcing a clean separation of concerns between qualitative AI reasoning and deterministic risk controls. Built natively on Alpaca's API ecosystem, the agent operates in five distinct stages: Quantitative Screener: Queries Alpaca market data to track S&P 100 leaders down over 15% from 52-week highs. Qualitative LLM Triage: Ingests real-time headlines and filings via the Alpaca News API. An LLM agent powered by Groq and Pydantic classifies each selloff as either a Transitory Dip or Structural Failure, validating moat durability with structured JSON schemas. Options Structuring Engine: Transforms validated theses into 20–60 DTE Bull Put Credit Spreads, capturing elevated panic premium while buying lower put wings for disaster insurance. Deterministic Risk Gate: The LLM is strictly prohibited from sizing capital. Pure Python code calculates contract sizing to enforce a hard trade risk cap of ≤ 2.0% of total account equity, rejecting unhedged exposure. Execution & Observability: Dispatches atomic multi-leg orders to Alpaca Paper Trading, monitored via an interactive Streamlit UI displaying Greeks, conviction scores, and live executions. Phoenix Alpha bridges agentic news intelligence with unbreakable mathematical safeguards to deliver a production-grade autonomous trading desk.