TIME-DELAY MULTISCALE MULTIFRACTAL DETRENDED PARTIAL CROSS-CORRELATION ANALYSIS OF HIGH-FREQUENCY STOCK SERIES

Abstract

To study the time-delay multiscale partial cross-correlation between the high-frequency stock series SSEC and SZSE in the Chinese stock market, this paper proposes a new partial cross-correlation exponent, namely the [Formula: see text]-order time-delay multiscale partial cross-correlation exponent (denoted by [Formula: see text]). We integrate the time-delay factor and the multiscale range fluctuation function into partial cross-correlation analysis. Numerical experiments show that the newly proposed exponent [Formula: see text] can be used to detect the partial cross-correlation of two time series with time delay. After excluding the influence of other stock indexes, we found that the persistence and multifractal characteristics between SSEC and SZSE have become stronger. Among them, the American high-frequency stock series has a greater impact on the two Chinese stocks than the Asian high-frequency stock series. Comparing the methods of time-delay multiscale multifractal detrended partial cross-correlation analysis method (time-delay MM-DPXA) and time-delay multiscale multifractal detrended cross-correlation analysis method (time-delay MM-DCCA), we find that cross-correlation and partial cross-correlation have different properties on different time scales and different time delays.