Building optimal regime-switching portfolios

This paper introduces a novel portfolio optimization method, the Clustered Minimum Spanning Tree Nested Optimization, capable of overcoming the limitations of classical asset allocation, such as instability and over-concentration of portfolio weights, and providing a defensive mechanism against the enhanced systematic risk during high-volatility periods. To do so, we follow a graph theory and clustering-based multi-step approach that accounts also for volatility regime switches. In a bootstrapping setup, we show that our approach produces well-diversified and stable portfolios outperforming the competing methods in terms of risk-adjusted performance while curtailing tail risk by achieving lower portfolio kurtosis.