Portfolio Optimization Engine

Problem

Portfolio analytics written in notebooks rarely survives reuse. The calculations are sound but untested, uncached, and impossible to drive from anything but the notebook itself.

Approach

• 01Architected a modular Python engine with multi-asset ingestion across equities, indices, forex, and crypto via yfinance, with joblib-backed caching.
• 02Implemented risk and performance metrics including Sharpe ratio, VaR and CVaR at 95 percent, maximum drawdown, and Omega ratio.
• 03Developed a Monte Carlo framework using a multivariate normal with covariance fallback to model portfolio path distributions.
• 04Structured the codebase into separated source, simulation, calculation, and visualisation layers with pytest coverage across metrics and data pipelines.
• 05Designed a Tkinter GUI for interactive fetching, simulation, and risk visualisation.