Intelligent Soft Computing on Forex: Exchange Rates Forecasting with Hybrid Radial Basis Neural Network

Area of Science:

• Quantitative finance
• Computational intelligence
• Financial forecasting

Background:

• Accurate prediction models are crucial for management decision-making in finance.
• Existing models may lack the precision required for high-frequency financial data.
• Soft computing offers advanced techniques for complex prediction tasks.

Purpose of the Study:

• To develop and evaluate a novel hybrid neural network for financial time series prediction.
• To enhance the accuracy of exchange rate forecasting using a combination of soft computing techniques.
• To assess the model's performance against established forecasting methods.

Main Methods:

• Development of a hybrid neural network integrating Radial Basis Function (RBF) network, genetic algorithm, and moving average.
• Application of the model to high-frequency USD/CAD exchange rate time series data.
• Comparative analysis using out-of-sample testing against autoregressive models and standard neural networks.
• Optimization of Artificial Neural Network (ANN) parameters using genetic algorithms, compared to backpropagation and K-means clustering.

Main Results:

• The proposed hybrid neural network demonstrated superior accuracy in forecasting exchange rate values.
• The model outperformed standard autoregressive and neural network models in one-day horizon predictions.
• Genetic algorithm optimization proved effective in enhancing ANN parameter adaptation.
• The hybrid model showed potential in reducing decision-making risks in financial management.

Conclusions:

• The hybrid neural network offers a more accurate prediction tool for financial markets.
• This approach effectively leverages soft computing for improved financial forecasting.
• The model provides valuable insights for informed management decisions in volatile financial environments.