Skip to content

PHASES.md

Latest commit

 

History

History
41 lines (27 loc) · 1.85 KB

PHASES.md

File metadata and controls

41 lines (27 loc) · 1.85 KB

🛠️ Project Evolution & Technical Milestones

This document tracks the iterative development of the Stock Risk Engine, from initial data ingestion to institutional-grade model validation.

✅ Phase VI: Macro Stress Testing & AI Governance (March 2026)

Goal: Implement systemic risk simulations and align with NIST AI RMF standards.

  • Stress Simulation Engine: Developed a 'What-If' module using historical correlation matrices (GFC 2008, COVID 2020, Tech Bubble 2000).
  • Predictive Risk Mapping: Integrated "Beta Shift" logic to account for correlation convergence during liquidity crises.
  • AI Governance Layer:
    • Implemented Model Cards for transparency.
    • Aligned documentation with ISO/IEC 42001 and NIST AI RMF.
    • Automated Violation Rate monitoring (Current: 2.47%).
  • UI/UX Refinement: Moved global parameters to the sidebar and implemented sub-tab navigation for "Historical Audit" vs. "Predictive Simulation."

Phase V: Model Validation & Governance

  • Goal: Certify model accuracy.
  • Outcome: Achieved 3.28% violation rate.
  • Innovation: Migrated backtesting to the internal Silver Layer for zero-lag reporting.

Phase IV: The Gold Layer & Visualization

  • Goal: Curate data for high-performance reporting.
  • Outcome: Built a Streamlit dashboard with 10,000-path Monte Carlo simulations.

Phase III: Risk Engineering Core

  • Goal: Quantify downside risk.
  • Outcome: Automated 95% VaR calculations using a 130-day trailing window.

Phase II: The Silver Layer (Feature Engineering)

  • Goal: Transform raw prices into actionable metrics.
  • Outcome: Developed automated pipelines for Rolling Volatility and Beta.

Phase I: The Bronze Layer (Data Ingestion)

  • Goal: Build the foundation.
  • Outcome: Designed a resilient SQLite schema and yfinance scraper.