Assessing the Performance of Prediction Models for the KSE 100 Index: Empirical Evidence based on ARIMA, Artificial Neural Networks, and Hybrid Model

Abstract

The prediction of the stock market has paved the way for minimizing the awareness of risk perception associated with portfolio investment. Previous studies evaluated multiple methods to find the best predictive model for future stock prices but concluded that there is no single model suitable for all stock markets. Consequently, for the investor, there is a great need to find the best predictive model for a rational investment decision. In this context, several models such as autoregressive integrated moving average (ARIMA), exponential smoothing, artificial neural networks (ANN), and hybrid (neural network and ARIMA, ARIMA and exponential smoothing, and neural network and exponential smoothing) are tested for the Pakistan Stock Market for the daily closing, open, high, and low price of the KSE 100 Index from 2000 to 2022 to explore the best predictive model. The results show that ARIMA is the best method for the prediction of stock prices based on the KSE 100 index and suggest that the stakeholder follows the pattern predicted by ARIMA.