This study aimed to evaluate the ability of ARIMA-ARCH and Neural Networks models to forecast future movement in Damascus Securities Exchange (DSE). In order to accurately compare the two methods in predicting the direction of the DSE movement, the study data were divided equally (From 1/1/2018 to 18/8/2019) assigned to build the models and to train the neural networks. (From 19/08/2019 to 19/09/2019) assigned to forecast the movement of DSE Index for both methods. In order to achieve fair comparison between the ability of neural networks with the versions of the ARIMA-ARCH models, The previous values of the studied variable (the closing price of the index) were solely relied upon as inputs to the artificial neural network to predict the next day without calculating any technical indicators and without introducing any other variables that help predict the values. The network coefficients (inputs - number of neurons) was set using a trial and error rule. Two neural networks have been proposed that can be used to predict the movement of the DSE index. One of the most important results is that; the most suitable model for predicting the DSE using ARCH-ARIMA models is ARIMA (1.1.1) and ARCH (1). It is not possible to rely on the models of ARCH-ARIMA to predict the direction of the movement of the DSE index, while this can be done using artificial neural networks and a compatibility proportion up to 78% according to the predicted data. Thus, it is possible to rely on artificial neural networks to make accurate investment decisions.
Hama University Journal.
2019.
Students
Teachers
Graduates
Faculties
