The Study of Hidden Markov Model for Estimating Stocks Index Volatility

Abstract

Volatility is a measure of uncertainty, which is useful for investor to plan a good investment strategy. The problem is that volatility is unobservable, and estimating volatility is not a trivial task. Hidden Markov model is, a model which is consisted of two processes, i.e. observation and a Markov process. The Markov process is assumed to be unobserved (hidden). In this thesis, LQ45 index is considered as the observation process and the unobserved hidden Markov is the volatility. To estimate the volatility of LQ45 index, the model proposed by Rossi and Gallo (2006) is used. The result of the study shows that volatility estimation of the LQ45 index using the model performs well. This is shown by the calculated error using symmetric mean absolute percentage error, which is only about 13.62%.