Developing new portfolio strategies by aggregation

We propose a method to combine N portfolio strategies by optimizing a given utility function $$U(\cdot )$$U(·). The method does not rely on any distributional assumption, could be easily extended to different combining functions and does not require any closed-form solution for the portfolio strategies to be combined. By focusing on three utility functions and a pool of five portfolio strategies, empirical analyses on real-world data show that the new method allows us to build combinations that better exploit the strengths of the different portfolio strategies during different market periods, thereby adapting to the data at hand and often outperforming state-of-art benchmarks.