Dynamic Risk Allocation with Carry, Value and Momentum

Abstract

According to recent research, diversification across risk factors (or investment styles) proves to be more efficient than traditional asset class diversification. In this paper, we take the next step and show that it is economically worthwhile to combine risk factors in a dynamic manner, in a process that we call Dynamic Risk Allocation (DRA). Building a DRA portfolio by means of several unconventional heuristics adds robustness and intuition to the whole portfolio construction process.Our main finding is that risk factor allocation largely replaces traditional global equity and bond market premiums as well as allocation to hedge funds (in expected utility maximization sense). Hence we question the economic validity of the alpha-beta separation paradigm that currently prevails in the industry. We show that our results are robust to transaction costs and portfolio constraints. We analyse a global portfolio of 3 well-known risk factors: momentum, value and carry. Each risk factor is broadly diversified across 4 global asset classes and taken both in cross-sectional and time series contexts. We test our approach using 38 years of daily historical prices in a broad set of futures contracts and major FX rates.