Mixed frequency data and portfolio selection: A novel approach integrating DEA with mixed frequency data sources

Abstract

This paper presents an innovative approach to portfolio optimization by integrating key elements of asset selection, risk management, and portfolio rebalancing. We first employ the Mixed Data Sampling (MIDAS) model to accurately measure Expected Shortfall (ES). Then, the Range Directional Measure-based Data Envelopment Analysis is considered to assess the portfolio efficiency, which integrates ES, asset returns, and inter-asset correlations for asset selection. Finally, utilizing the mixed frequency data from the metal futures market, we compared the portfolio performance of the Global Minimum ES strategy and the Market Neutral strategy, which reveals that our framework always outperforms traditional benchmarks in multiple aspects. Our findings indicate that, under the comprehensive risk management, a weekly rebalancing strategy is more effective compared to a daily rebalancing scheme. Furthermore, our study demonstrates that stringent asset selection, as opposed to loose selection or non-selection, significantly enhances the overall portfolio performance under the comprehensive risk management. Collectively, this research underscores the necessity of judicious asset selection and rebalance strategies in the modern portfolio management, and validates the practical utility of the portfolio efficiency with DEA and the mixed frequency data sources with MIDAS scheme.