Determining Hidden States in selected Stock Indices using Hidden Markov Models

Authors

  • P.V.Chandrika, K. Sakthi Srinivasan

Abstract

The present study aims at exploring hidden states and true regimes in selected stock indices
S&P 500 and Dow Jones Index. Hidden Markov Models (HMM) with Expectation
Maximization algorithm is applied to find out true regimes in the stock indices. The closing
price of the stock index is given as the observed states to HMM model to determine unobserved
states. The data considered for the study is from Jan 2014 to May 2020. The results show that
even though the algorithm is trained taking posterior probabilities of determining higher states,
the stock market clearly exhibits two states of bullish and bearish markets which are the general
states in stock market.

Published

2020-10-03

Issue

Section

Articles