.png&w=256&q=75)
🤓 Latest Submissions
Sentinel AI
Sentinel is an adversarial AI-powered options trading system designed to make automated trading more disciplined, transparent, and risk-aware. Instead of allowing a single AI model to generate and execute a trade, Sentinel uses a multi-stage decision pipeline: Market Intelligence → Strategy Generation → Adversarial Challenge → Risk Validation → Execution. The system first analyzes market conditions and develops a directional thesis. A strategy agent then constructs an options position with defined strikes, maximum loss, maximum profit, and rationale. Before execution, an independent adversarial agent actively red-teams the strategy, searching for weaknesses, unfavorable assumptions, and market risks. A deterministic Risk Engine then evaluates hard constraints including maximum loss, position size, daily trade limits, and portfolio exposure. Only strategies that pass these safeguards can proceed to execution. Sentinel integrates Alpaca's MCP Server / CLI and Trading API and operates against a fresh $100,000 paper-trading portfolio for standardized evaluation of risk, drawdowns, and performance. Every decision stage is persisted as structured JSON audit logs, including market analysis, trade thesis, adversarial findings, risk checks, and execution results. The dashboard provides a complete, transparent view of the decision trail. The AI proposes. The adversary challenges. Risk controls decide. Alpaca executes.
4 Sep 2026
