AlphaPilot - Autonomous Options Agent

AlphaPilot is an autonomous AI trading agent designed to identify, evaluate, execute, and manage options trades in Alpaca paper trading. Instead of relying solely on an LLM or a traditional quantitative strategy, AlphaPilot combines both approaches. A quantitative strategy continuously scans the market for potential opportunities using market conditions, momentum, volatility, and options-specific signals. Candidate trades are then passed to an AI supervisor that evaluates the setup, market context, thesis, confidence, and potential risks. The AI does not have unrestricted control over execution. Every proposed trade must pass through a deterministic risk engine that acts as the final authority. Position sizing, portfolio exposure, maximum positions, daily loss limits, drawdown limits, concentration limits, and execution constraints are enforced independently of the AI's decision. The agent operates through an autonomous loop: scan the market → calculate features → generate signals → evaluate with AI → validate through risk controls → execute → monitor positions → repeat. AlphaPilot also maintains a decision journal that records the reasoning behind each trade, allowing users to trace a decision from the original market signal through AI analysis, risk validation, execution, and outcome. The project includes a control dashboard where the agent can be started, stopped, monitored, and configured. Users can inspect live portfolio performance, open positions, trades, risk exposure, agent activity, and historical decisions from a single interface. The goal is not simply to build an AI that can generate trading ideas, but to demonstrate a complete, testable autonomous trading system where AI reasoning is combined with quantitative signals, deterministic safety controls, transparent decision-making, and continuous paper-trading execution.