Enhancing Volatility Forecasting : A Dual Empirical Mode Decomposition Approach

The paper aims to enhance the accuracy of realized volatility prediction by introducing a novel Dual Empirical Mode Decomposition (DEMD) method that allows for the extraction of incremental information related to volatility prediction in raw financial data. The empirical results show that using the DEMD method to decompose and reconstruct trading volume leads to a demonstration of superior in-sample explanatory power. Furthermore, in terms of out-of-sample volatility forecasting, the DEMD method exhibits significant advantages over the direct addition of trading volume and the use of similar methods employed in popular volatility and linear regularization models