Memory Effect Modelled By Two-State Markov Process – Analysis Of Selected Markets

Abstract:

In the paper there are presented results of research on the occurrence of memory effect (in the time series of rates of return of the analysed companies) modelled by a two-state Markov process. In the described model the state is determined by the sign of the last observed rate of return, and the model itself distinguished two states, defined as the "negative" state (when the last observed rate of return was negative) and the "positive" state (where the last observed rate was positive), the cases with rates of return zero were omitted. In each of the two states the variable that was the rate of return could come from a different probability distribution. If there were significant differences between the two probability distributions, it was assumed that memory effect occurred.

The research was conducted on instruments listed on the stock exchanges: the New York Stock Exchange, the London Stock Exchange, the Stock Exchange of Hong Kong in time period  01.05.2016 - 31.04.2021. The results of the research show that the memory effect was observed in some of the studied instruments.