Multi-Period Portfolio Optimisation Using a Regime-Switching Predictive Framework

TL;DR

This paper introduces a novel multi-period portfolio optimization framework that combines regime prediction with model predictive control, utilizing a Kalman filter to enhance return estimates, and demonstrates superior performance over standard benchmarks.

Contribution

The paper develops the KMRF+MPC model integrating regime prediction with portfolio optimization, improving return estimation accuracy and outperforming industry benchmarks.

Findings

Outperforms industry-standard benchmarks in portfolio returns.

Uses Kalman filter to enhance regime prediction accuracy.

Applicable to long-only investment strategies.

Abstract

Regime-switching poses both problems and opportunities for portfolio managers. If a switch in the behaviour of the markets is not quickly detected it can be a source of loss, since previous trading positions may be inappropriate in the new regime. However, if a regime-switch can be detected quickly, and especially if it can be predicted ahead of time, these changes in market behaviour can instead be a source of substantial profit. The work of this paper builds on two previous works by the authors, the first of these dealing with regime detection and the second, which is an extension of the first, with regime prediction. Specifically, this work uses our previous regime-prediction model (KMRF) within a framework of multi-period portfolio optimisation, achieved by model predictive control, (MPC), with the KMRF-derived return estimates accuracy-boosted by means of a novel use of a Kalman filter. The resulting proposed model, which we term the KMRF+MPC model, to reflect its constituent methodologies, is demonstrated to outperform industry-standard benchmarks, even though it is restricted, in order to be acceptable to the widest range of investors, to long-only positions.