Analysis of Multifractal Cross-Correlation Characteristics and Information Flow of Typical Stock Prices in the U.S. Industry Sectors

This paper investigates multifractal cross-correlations and information flow among daily closing prices of stock-typical U.S. industry sectors, combining multifractal detrended cross-correlation analysis (MFDCCA), transfer entropy (TE), and complex network methods. Firstly, the ensemble empirical mode decomposition (EEMD) method estimates the dominant frequency. Then, the DCCA coefficient is computed, revealing that JPM exhibits nonlinear cross-correlations with all seven other stocks. Secondly, a more detailed examination using MFDCCA elucidates fractal characteristics. The experimental results reveal long-range and multiple fractal characteristics in the cross-correlations between JPM and the remaining seven stocks. Notably, the strongest multifractal cross-correlation is observed between JPM and XOM, while the weakest is between JPM and SPG. Thirdly, transfer entropy is calculated for each pair of the eight stocks to research the direction of information flow. The analysis reveals bidirectional information transfers, which are notable for the high information transfer from PG to XOM, as indicated by the transfer entropy matrix. Finally, utilizing a complex network approach to visualize the transfer entropy results, it is evident that AAPL possesses the most significant information outflow, while XOM exhibits the most substantial information inflow. These findings present critical insights beneficial for portfolio decision-making in the stock markets.