Modeling Stock Index Using Finite State Markov Chain

Existing stock price models are based on time series methodologies which are hard to estimate and involves lots of assumptions. This study, in contrast, assumes that the stock prices follow stochastic process that possesses Markov dependency with finite state transition probabilities. For this purpose, daily stock index data from Pakistan Stock Exchange (PSE) is collected from 2010-2015 and categorized in to 10 state spaces. Based on the results of state transition model, the study highlights the most probable state of return and also its transition into another state. Further, the study used Monte Carlo method of stock index simulations both Markov chain and original stock index. The analysis shows that it is possible to model and forecast stock index by capturing Markov process. The results of the study are helpful for investors in selecting right time of making investment and for academician to think about more sophisticated methods of state identification.