Application of Artificial Neural Networks (ANN) in Forecasting the Dubai Financial Market Index

Abstract

Accurate forecasting of the Dubai Financial Market (DFM) index constitutes a strategic tool for supporting investors' and portfolio managers' decision-making processes, given the volatile and nonlinear nature of financial markets.

  This study aims to evaluate the efficiency of neural network models in forecasting the daily closing values of the DFM index, based on historical financial data.

  MATLAB R2023a was utilized to test several neural predictive models with varying architectures, in order to identify the most optimal model in terms of accuracy and predictive performance. The best-performing model employed a Logsig activation function in the hidden layer and a linear function in the output layer.

  The results of the selected model demonstrated high accuracy, with a Mean Squared Error (MSE) of approximately 97.0, a Mean Absolute Error (MAE) of 7.68, a Mean Absolute Percentage Error (MAPE) of 0.21, and a Coefficient of Determination (R²) of around 0.9989. These metrics reflect the model's strong capability in capturing the underlying relationships between variables.

  The findings suggest that neural network models are effective in analyzing financial time series and extracting complex patterns, thereby enhancing their value as decision support tools in dynamic market environments.