Purpose: The research aims to compare the results of forecasting financial market indicators in the Arabian Gulf using the hybrid ARIMA-GARCH model. This research argues for the efficiency of traditional standard models in forecasting for relatively long periods. Methodology: The hybridization technique initially depends on the selection of the best autoregressive integrated moving average model for prediction according to the automatic selection function for the proposed models that achieve the best statistical conditions. The models chosen in the previous step represent the inputs for using the (1,1,1) GARCH model for the prediction. Both models predict financial market indicators from outside the sample for 12 months. Finally, the actual closing points of the market indicators and the prediction results of the two models were compared. Findings: The research concludes that ARIMA models are effective and efficient in predicting the indices of the Kuwait Stock Exchange and the Manama Stock Exchange. By contrast, the GARCH model outperformed the ARIMA model in predicting the Saudi Stock Exchange index and achieved excellent results. The prediction results for the Iraqi and Dubai markets did not achieve satisfactory results; however, the results of the comparison by calculating the average square error indicate the superiority of the ARIMA models. Originality: This study argues the possibility of traditional standard models in predicting financial market indicators for relatively long periods, which is not provided by previous literature that argues the efficiency of artificial intelligence models and confirms that standard models can predict relatively short periods.
Adnan et al. (Fri,) studied this question.