Testing the Stochastic Disorder Model on Stock Markets

Abstract

This paper examines real-time applications of quickest disorder detection techniques for timing stock markets. The focus is on the stochastic disorder model by Shiryaev, Zhitlukhin, and Ziemba (2014, 2015), Zhitlukhin and Ziemba (2016) and their optimal stopping rule. The model uses sequential price data to identify a directional change in the market trend and determines the optimal exit moment from a long position in a bubble-like market. Together with the sensitivity analysis of the exit rule's signals to model parameters, we study out-of-sample performance of the entry-exit investment strategy that exploits signals from the rule. Using historical data on the S&P 500, we find that the entry-exit strategy underperforms the buy-and-hold strategy over the whole testing period 1975-2016, but outperforms it in the fall of 1987 and during the bear market of 2007-2009.