Abstract:
In this paper, we employ a four-state hidden semi-Markov model, which outperforms a hidden Markov model, to identify market conditions of the US stock market over the daily period from 16th of February, 1885 to 4th of June, 2020. Our results indicate that the four hidden states represent bear-, bull-, sidewalk-, and crash-markets, which in turn appropriately capture the various major historical events during the period of study.