Financial Time Series Prediction Using Elman Recurrent Random Neural Networks

Area of Science:

• Economics
• Computer Science
• Data Science

Background:

• Financial market dynamics forecasting is a critical area of economic research.
• Accurate prediction of stock market price indices is essential for investment strategies.

Purpose of the Study:

• To develop and evaluate a novel architecture for enhanced financial time series forecasting.
• To assess the predictive capabilities of the proposed model across various international stock market indices.

Main Methods:

• Developed a hybrid model combining Elman recurrent neural networks with a stochastic time effective function.
• Utilized linear regression, complexity invariant distance (CID), and multiscale CID (MCID) for model analysis.
• Compared the proposed model against Backpropagation Neural Network (BPNN), Stochastic Time Effective Neural Network (STNN), and Elman Recurrent Neural Network (ERNN).

Main Results:

• The proposed neural network architecture demonstrated superior performance in financial time series forecasting compared to benchmark models.
• Empirical testing on SSE, TWSE, KOSPI, and Nikkei225 stock market indices confirmed the model's strong predictive accuracy.
• Statistical comparisons validated the effectiveness of the developed approach for predicting stock market index values.

Conclusions:

• The novel Elman recurrent neural network combined with stochastic time effective function offers a robust solution for financial market forecasting.
• The model's effectiveness is confirmed across diverse Asian stock markets, highlighting its practical applicability.
• This research contributes a valuable tool for improving the accuracy of stock market predictions.