🤓 Latest Submissions
Magno
Autonomous AI agents are shifting from passive market analysis to active trade execution. However, unconstrained LLM trading systems often suffer from execution drift, toxic bid-ask spread slippage, and unhedged directional drawdowns. Magno solves this by combining Alpaca's Paper Trading API, Featherless AI, the Model Context Protocol (MCP), and nine deterministic quantitative risk gates into an institutional-grade autonomous options trading desk. Key Architecture and Capabilities: Closed-Menu Strategy Discovery: Powered by Featherless AI (Qwen2.5-72B-Instruct), Magno scans liquid assets (SPY, QQQ, NVDA) using Alpaca options chains. Operating within a strict schema, the reasoner evaluates IV rank against realized variance without hallucinating strikes or parameters. 9 Deterministic Python Risk Gates: Before routing orders, the engine enforces non-negotiable checks including a hard 5% maximum bid-ask spread ceiling, margin allocation caps, DTE limits, and defined-risk vertical spread enforcement. Autonomous Dynamic Delta Hedging: The engine monitors portfolio Greeks in a continuous 5-second loop. Whenever exposure drifts past threshold, Magno autonomously dispatches fractional underlying equity orders on Alpaca to snap exposure back to 0.00 Delta. Local Newton-Raphson Greeks Engine: Implied volatility and Greeks are computed locally on-device, eliminating reliance on delayed broker feeds while isolating Greek buckets per underlying asset. Native FastMCP Server Layer: Exposes standardized tools for querying telemetry, screening volatility, and dispatching risk-gated orders across external agent workflows.
4 Sep 2026
