Risk-return adaptive receding Horizon Index Tracking Strategy

Abstract

Index tracking is a well-established financial strategy for passive investing. Typical index tracking models are single period in nature, deriving an optimal tracking portfolio based on future price/return estimates, using most if not all index constituent assets. In this article, we propose a framework for index tracking that can accommodate multi-periods and asset selection. First, we propose a risk-return-based index tracking strategy within a multi-period adaptive receding horizon framework. The framework demonstrates strong tracking fidelity with the benchmark whilst accounting for future tracking states. We then adapt a Penalized Alternating Direction Method (PADM) to the multi-period framework to efficiently enforce a limit on tracking portfolio size (cardinality). The PADM produces high-quality solutions to the cardinality-constrained models and can be used effectively in both low and higher re-balancing frequency environments. Finally, we generalize our base multi-period formulation to an enhanced index tracking strategy, which can easily accommodate possible portfolio manager (PM) preferences. We present computational results that indicate that our cardinality-constrained and non-cardinality-constrained adaptive receding horizon framework for index tracking yields high tracking accuracy when compared to equivalent single-period or return-based models used in a rolling horizon framework.