Using Markov-Switching models in Italian, British, U.S. and Mexican equity portfolios: a performance test

Abstract

In this paper we test the use of Markov Switching models in equity trading strategies, following Brooks and Persand (2001), Kritzman et al. (2012) and Hauptmann et al. (2014), who suggest their use as warning systems of bad performing periods. We extend their reviews by testing again (with the impact of trading fees) the U.S. and U.K. markets and by extending our tests to the Italian and Mexican case. The rationale behind our Markov-Switching strategy is to invest in equity index tracking ETFs in low volatility or ”good performing” periods and in the local risk-free asset in high-volatility or ”bad performing” ones. Our results show that in a weekly simulation from January 4, 2001 to July 30, 2017 with a 0.35% trading fee plus taxes, our system is useful to create alpha in all the simulated markets even if the Italian case showed several deep distress moments due to a financial or political crisis.