Timing for portfolio rebalancing: a data-driven robust optimization framework

Abstract

In portfolio management, the time spans between successive rebalancing points are a key decision. We present a robust optimization framework to manage the timing of rebalancing points in the context of discrete-time portfolio rebalancing. The framework is a variation of the calendar balancing approach. For a given number of rebalancing points in the investment horizon, time periods between successive rebalancing time points are modeled as decision variables. Rebalancing time points that may potentially yield higher returns are generated with a robust optimization model that processes realized returns. The optimization model is a network flow problem with side constraints and can be solved with minimal computational effort. We evaluate the out-of-sample performance improvements from using the robust optimization framework vis-& agrave;-vis the quarterly rebalancing strategy, over the 1971-2020 period. The computational analysis indicates that statistically significant return improvements can be attained with the proposed robust optimization framework.