Improving forecasting accuracy for stock market data using EMD-HW bagging.

Area of Science:

• Quantitative Finance
• Time Series Analysis
• Computational Economics

Background:

• Stock market data are characterized by nonstationary and nonlinear properties, posing significant challenges for traditional forecasting models.
• Accurate forecasting of stock market trends is crucial for investment strategies and economic stability.

Purpose of the Study:

• To develop and evaluate a novel forecasting method for nonstationary and nonlinear time series data, specifically applied to stock market data.
• To assess the performance of the proposed Empirical Mode Decomposition-Holt-Winter (EMD-HW) bagging method against existing forecasting techniques.

Main Methods:

• The study employs the Empirical Mode Decomposition (EMD) to decompose the time series into intrinsic mode functions.
• The Holt-Winter (HW) method is integrated with a moving block bootstrap approach within a bagging framework (EMD-HW bagging).
• The EMD-HW bagging method was applied to the stock market time series data of six different countries.

Main Results:

• The EMD-HW bagging method was compared against fourteen other selected forecasting methods.
• Performance was evaluated using five distinct forecasting error measurements.
• Results indicated that the EMD-HW bagging method achieved higher accuracy in forecasting stock market time series compared to the benchmark methods.

Conclusions:

• The EMD-HW bagging method is a robust and accurate approach for forecasting nonstationary and nonlinear time series, particularly in the context of stock markets.
• This method offers a significant improvement over existing techniques, providing more reliable predictions for financial market data.