Quantitative Portfolio Management Strategy Based on the Combination of Mean-Variance Optimization and Time-Series Momentum

Abstract

As financial market complexity increases, the need for effective and reliable quantitative portfolio management strategies has become more pressing. This study integrates Mean-Variance Optimization (MVO) with Time-Series Momentum (TSM) to create a hybrid strategy aimed at dynamically adjusting portfolio weights and improving overall performance. By leveraging historical stock data, the strategy was rigorously evaluated through in-sample testing (2010–2019) and out-of-sample testing (2019–2024). The findings reveal remarkable enhancements in cumulative returns, achieving 800% in-sample and 650% out-of-sample, underscoring the approach’s robustness across varying market conditions. Additionally, risk metrics illustrate a delicate balance between long-term stability and short-term adaptability, offering insights into the strategy's effectiveness. However, elevated drawdowns and increased volatility in the out-of-sample phase raise concerns about potential overfitting and warrant further refinement. This study highlights the potential of combining MVO and TSM to enhance portfolio management by capitalizing on market trends while maintaining risk control. Future research could explore the incorporation of machine learning algorithms or alternative momentum signals to optimize performance further and address the challenges of adapting to diverse and evolving market environments.