Enhancing risk parity models: a two-stage approach using mean-variance and expected shortfall for optimal asset selection

Abstract

In a progressively complex and volatile financial scenery, the need for robust investment strategies has never been greater. Portfolio optimization seeks to balance risk and return for superior performance. Traditional approaches, such as the Mean-Variance model by Markowitz [1952], laid the groundwork for diversification and risk management. Risk Parity models offer a compelling alternative by equalizing risk contributions across assets, but often lack an optimal mechanism for asset selection, leading to suboptimal results, especially with large and diverse asset universes. This paper proposes a novel two-stage approach. First, we use the Mean-Variance model and
Expected Shortfall (ES) to select a refined subset of assets, minimizing risk without imposing return constraints. Second, we apply Risk Parity techniques—standard deviation for the Mean-Variance subset, ES for the ES-selected subset—to balance risk contributions. Tested on both Nikkei 225 equities and a mixed portfolio of stocks, bonds, and cryptocurrencies, our method enhances resilience while preserving returns.