Assessing the predictability of all share price index of Colombo stock exchange using different models : a case study during the COVID - 19 pandemic

Abstract

The aim of this investigation was to assess the predictability of three models: Autoregressive Integrated Moving Average (ARIMA), Seasonal Auto-regressive Integrated Moving Average (SARIMA), and Dynamic Harmonic Regression (DHR) model, both prior to and following the Covid-19 outbreak. Every model was crafted with great care and then compared to determine the optimal method for predicting future outcomes. The findings suggested that, during the Covid-19 period, the DHR model outperformed the other models as it had the lowest Corrected Akaike’s Information Criterion (AIC) value. According to the Portmanteau test, the residuals were random and not correlated, indicating that all the models were adequate for making predictions. Although the rapid decline of CSE was captured by both the ARIMA and DHR models, the DHR model yielded more significant results. In contrast, prior to the pandemic, the ARIMA model performed well and effectively captured the underlying trend compared to other models. However, forecast errors indicated that DHR model was more appropriate for predicting daily share indices with long intricate seasonal variations compared to the SARIMA model. As a consequence, stakeholders were able to make accurate investment decisions even in the midst of the outbreak. Finally, the Engle’s ARCH test was conducted to analyze the occurrence of volatility clusters during the pandemic, and it was identified that there were notable fluctuations in volatility throughout the pandemic period.