Comparative study of the Application of Seasonal ARIMA and Exponential Smoothing Methods in Nigerian Stock Exchange Market

Abstract

This paper compared the performance of two forecasting models (Seasonal ARIMA and 
Exponential smoothing) in an attempt to identify the model that fits properly in forecasting 
Nigerian stock exchange market. A two-staged approach to forecasting was carried out using 
monthly data for the period of 1985 to 2013. The models were assessed in similarly structured 
setting at the beginning, and then best models identified at this level were compared in a 
differently structured setting. The results show that Seasonal ARIMA (4,1,3)(3,1,2)12 and 
Holt-Winters multiplicative smoothing method are effective in forecasting Nigerian stock 
exchange market in a similarly structured setting. Nonetheless, when the two models were 
compared under different structures, the performance of Holt-Winters multiplicative 
smoothing method outperformed that of Seasonal ARIMA (4,1,3)(3,1,2)12. This suggests that 
Holt-Winters multiplicative smoothing method with Alpha (0.01), Delta (0.11) and Gamma 
(0.11) is more effective in forecasting Nigerian stock exchange market in the short run and it 
can be used to aid planning processes in the stock exchange market. Likewise, the seasonality 
pattern that characterizes stock exchange highlights the need to promote more of stock 
exchange market so as to lessen the negative impacts associated with it. The two models can 
be adequately used to forecast stock exchange data as the results have shown their potentiality 
in that regard.