Volatility Targeting Portfolios: A Multi-Period Framework

Abstract

Numerous empirical as well as theoretical studies have demonstrated the superior performance of dynamic asset allocation strategies that target a constant volatility level through time. These volatility targeting strategies mitigate portfolio risk by adjusting the portfolio’s exposure based on updated volatility forecasts. However, most dynamic allocation strategies use a myopic optimization scheme; namely, they repeatedly solve a sequence of single period optimization problems. This paper will investigate the use of a multi-period decision model in the construction of volatility targeting portfolios. Counter to the predominant stochastic programming approach, which is hindered by its computational intractability, this thesis uses an optimization scheme that can be solved by readily available convex quadratic programming solvers. Exploring two different recourse policies, this paper finds that the multi-period approach outperforms the myopic scheme when measured on a risk-return basis.