Hybrid Quantum-Classical Portfolio Optimization

Optimizing S&P 500 returns using LightGBM and Quantum Approximate Optimization.

Research Overview

As a Graduate Researcher, I developed a system to optimize portfolios for S&P 500 stocks. [cite_start]The system processes over 750k+ rows of data, reducing reload times by 5x using Parquet caching[cite: 20].

Methodologies

• [cite_start]Machine Learning: Utilized LightGBM on momentum and NLP sentiment data to predict expected returns[cite: 21].
• [cite_start]Risk Management: Implemented Black-Litterman models and Mean-CVaR optimization[cite: 21, 22].
• [cite_start]Quantum Extension: Extended the framework to use VQE and QUBO (Quantum Unconstrained Binary Optimization) for accelerated estimation[cite: 23].

Results

• [cite_start]Achieved a 20% reduction in downside risk in backtests[cite: 22].
• [cite_start]Maintained a Sharpe ratio > 1 through walk-forward validation[cite: 22].