Forecasting stochastic neural network based on financial empirical mode decomposition.

Area of Science:

• Computational Finance
• Time Series Analysis
• Machine Learning

Background:

• Accurate stock price forecasting is crucial for financial markets.
• Traditional models often struggle with the complex, non-linear dynamics of stock prices.
• Existing methods may not fully capture the temporal dependencies and oscillatory modes within financial data.

Purpose of the Study:

• To develop a novel one-step-ahead forecasting model to enhance stock price fluctuation prediction accuracy.
• To integrate Empirical Mode Decomposition (EMD) with a Stochastic Time Strength Neural Network (STNN) for improved financial time series analysis.
• To validate the proposed model's effectiveness against traditional forecasting techniques.

Main Methods:

• Empirical Mode Decomposition (EMD) was employed to decompose financial time series into intrinsic oscillatory modes.
• A Stochastic Time Strength Neural Network (STNN) was developed, incorporating the temporal weight of historical data.
• The combined EMD-STNN model's predictive performance was assessed using linear regression and a q-order multiscale complexity invariant distance metric.

Main Results:

• The EMD-STNN model demonstrated superior accuracy in forecasting stock price fluctuations compared to traditional models.
• Empirical results confirmed the model's effectiveness in capturing complex market dynamics.
• The q-order multiscale complexity invariant distance metric validated the model's robust predictive capabilities on real stock index data.

Conclusions:

• The proposed EMD-STNN model offers a significant improvement in stock market forecasting accuracy.
• The integration of EMD and STNN effectively addresses the challenges of non-linearity and temporal dependencies in financial data.
• This advanced model provides a valuable tool for predicting stock market fluctuations.