Stock Market Price Forecasting using ARIMA vs ANN; A Case study from CSE

Abstract

Stock market prediction or forecasting is a challenging task to predict the upcoming stock values. Stock prices are nonstationary and highly noisy because stock markets are affected by a variety of factors. Traditionally, the next lag of time series is effectively forecast by a variety of techniques like Simple Exponential Smoothing, ARIMA. In particular, ARIMA has shown its success in accuracy and precision in predicting the next time-series lags. As part of the literature, very few studies have focused on Colombo Stock Exchange (CSE) to find new predictive approaches for the forecasting of high volatility stock price indexes. Different statistical approaches and economic data strategies have been widely applied to define market price movements and trends and the trade volume levels in CSE over the last ten years. This article explores whether and how the newly developed deep learning algorithms for the projection of time series data, such as the Back Propagation Neural Network, are greater than traditional algorithms. The results show that Deep learning algorithms like BPNN outperform traditionally based algorithms like the model ARIMA. The MAE and MSE values relative to ARIMA and BPNN, which suggests BPNN 's superiority to ARIMA.