MULTISCALE DETRENDED CROSS-CORRELATION ANALYSIS OF STOCK MARKETS

Fractals ◽  
2014 ◽  
Vol 22 (04) ◽  
pp. 1450007 ◽  
Author(s):  
YI YIN ◽  
PENGJIAN SHANG
Keyword(s):  
Correlation Analysis ◽  
Stock Markets ◽  
Cross Correlation ◽  
Statistical Characteristics ◽  
Detailed Knowledge ◽  
Stock Indices ◽  
Regional Markets ◽  
Cross Correlation Analysis ◽  
Cross Correlations ◽  
The Cross

In this paper, we employ the detrended cross-correlation analysis (DCCA) to investigate the cross-correlations between different stock markets. We report the results of cross-correlated behaviors in US, Chinese and European stock markets in period 1997–2012 by using DCCA method. The DCCA shows the cross-correlated behaviors of intra-regional and inter-regional stock markets in the short and long term which display the similarities and differences of cross-correlated behaviors simply and roughly and the persistence of cross-correlated behaviors of fluctuations. Then, because of the limitation and inapplicability of DCCA method, we propose multiscale detrended cross-correlation analysis (MSDCCA) method to avoid "a priori" selecting the ranges of scales over which two coefficients of the classical DCCA method are identified, and employ MSDCCA to reanalyze these cross-correlations to exhibit some important details such as the existence and position of minimum, maximum and bimodal distribution which are lost if the scale structure is described by two coefficients only and essential differences and similarities in the scale structures of cross-correlation of intra-regional and inter-regional markets. More statistical characteristics of cross-correlation obtained by MSDCCA method help us to understand how two different stock markets influence each other and to analyze the influence from thus two inter-regional markets on the cross-correlation in detail, thus we get a richer and more detailed knowledge of the complex evolutions of dynamics of the cross-correlations between stock markets. The application of MSDCCA is important to promote our understanding of the internal mechanisms and structures of financial markets and helps to forecast the stock indices based on our current results demonstrated the cross-correlations between stock indices. We also discuss the MSDCCA methods of secant rolling window with different sizes and, lastly, provide some relevant implications and issue.

Multifractal Fluctuation Analysis of Correlations Between the Sector Stock Markets in China and the US

2020 ◽  
pp. 2150031
Author(s):  
You-Shuai Feng ◽  
Hong-Yong Wang
Keyword(s):  
Correlation Analysis ◽  
Stock Markets ◽  
Cross Correlation ◽  
Rapid Development ◽  
Fluctuation Analysis ◽  
Cross Correlation Analysis ◽  
The Us ◽  
Cross Correlations ◽  
The Cross ◽  
Stock Indexes

With the rapid development of economic globalization, the stock markets in China and the US are increasingly linked. The fluctuation features and cross-correlations of the two countries’ markets have attracted extensive attention from market investors and researchers. In this paper, the fractal analysis methods including multifractal asymmetric detrended cross-correlation analysis (MF-ADCCA) and coupled detrended cross-correlation analysis (CDCCA) are applied to explore the volatilities of CSI300 and SP500 sector stock indexes as well as the cross-correlations and coupling cross-correlations between the two corresponding sector stock indexes. The results show that the auto-correlations, cross-correlations and coupling cross-correlations have multifractal fluctuation characteristics, and that the cross-correlations are asymmetric. Additionally, the coupling cross-correlation strengths are distinct due to the different influence of long-range correlations and fat-tailed distribution. Further, the co-movement between China and the US sector stock markets is susceptible to external market factors such as major economic events and national policies.

Multiscale Multifractal Detrended Cross-Correlation Analysis of High-Frequency Financial Time Series

2019 ◽  
Vol 18 (03) ◽  
pp. 1950014 ◽  
Author(s):  
Jingjing Huang ◽  
Danlei Gu
Keyword(s):  
Time Series ◽  
Correlation Analysis ◽  
High Frequency ◽  
Cross Correlation ◽  
Financial Time Series ◽  
Chinese Market ◽  
Change Of Scale ◽  
Cross Correlation Analysis ◽  
Cross Correlations ◽  
The Cross

In order to obtain richer information on the cross-correlation properties between two time series, we introduce a method called multiscale multifractal detrended cross-correlation analysis (MM-DCCA). This method is based on the Hurst surface and can be used to study the non-linear relationship between two time series. By sweeping through all the scale ranges of the multifractal structure of the complex system, it can present more information than the multifractal detrended cross-correlation analysis (MF-DCCA). In this paper, we use the MM-DCCA method to study the cross-correlations between two sets of artificial data and two sets of 5[Formula: see text]min high-frequency stock data from home and abroad. They are SZSE and SSEC in the Chinese market, and DJI and NASDAQ in the US market. We use Hurst surface and Hurst exponential distribution histogram to analyze the research objects and find that SSEC, SZSE and DJI, NASDAQ all show multifractal properties and long-range cross-correlations. We find that the fluctuation of the Hurst surface is related to the positive and negative of [Formula: see text], the change of scale range, the difference of national system, and the length of time series. The results show that the MM-DCCA method can give more abundant information and more detailed dynamic processes.

Comparison of Multiscale Methods in the Stock Markets for Detrended Cross-correlation Analysis and Cross-sample Entropy

2014 ◽  
Vol 13 (03) ◽  
pp. 1450023 ◽  
Author(s):  
Yi Yin ◽  
Pengjian Shang
Keyword(s):  
Time Series ◽  
Correlation Analysis ◽  
Stock Markets ◽  
Cross Correlation ◽  
Sample Entropy ◽  
Multiscale Methods ◽  
Market Mechanisms ◽  
Cross Correlation Analysis ◽  
The Cross ◽  
The Difference

In this paper, we employ multiscale cross-sample entropy (MSCE), multiscale detrended cross-correlation analysis (MSDCCA) and DCCA cross-correlation coefficient (σDCCA) measurement to investigate the relationship between time series among different stock markets. We report the results of synchronism and cross-correlation behaviors in US and Chinese stock markets by these three methods. It can be concluded that the MSCE analysis point out the similarity on the cross-correlation among the stock markets while the MSCE makes it difficult to distinguish the indices in the same region and identify the difference and uniqueness of stock markets. However, both the MSDCCA analysis and σDCCA analysis reflect the similarity and uniqueness on the cross-correlation behaviors and reach the consistency. Furthermore, MSDCCA gives detailed multiscale cross-correlation structures and show some new interesting characteristics and conclusions, while the multiscale analysis by σDCCA provides a large amount of information on the cross-correlations and quantifies the level of cross-correlation more clearly and intuitively. MSDCCA and σDCCA methods may be more proper measures for the investigation of the cross-correlation between time series. We believe that such researches are relevant for a better understanding of the stock market mechanisms, and may lead to a better forecasting of the stock indices.

scholarly journals X-ray cross-correlation analysis of the low-mass X-ray binary 4U 1636-53

2014 ◽  
Vol 10 (S306) ◽  
pp. 397-399
Author(s):  
Ya-Juan Lei
Keyword(s):  
Correlation Function ◽  
Correlation Analysis ◽  
Cross Correlation ◽  
X Rays ◽  
X Ray ◽  
Cross Correlation Analysis ◽  
Low Mass ◽  
Cross Correlations ◽  
The Cross ◽  
Positive Correlations

AbstractWe analyze the cross-correlation function of the soft and hard X-rays of the atoll source 4U 1636-53 with RXTE data. The results show that the cross-correlations evolve along the different branches of the color-color diagram. At the lower left banana states, we have both positive and ambiguous correlations, and positive correlations are dominant for the lower banana and the upper banana states. The anti-correlation is detected at the top of the upper banana states. The cross-correlations of two atoll sources 4U 1735-44 and 4U 1608-52 have been studied in previous work, and the anti-correlations are detected at the lower left banana or the top of the upper banana states. Our results show that, in the 4U 1636-53, the distribution of the cross-correlations in the color-color diagram is similar to those of 4U 1735-44 and 4U 1608-52, and confirm further that the distribution of cross-correlations in color-color diagram could be correlated with the luminosity of the source.

DETRENDED CROSS-CORRELATION ANALYSIS BETWEEN MULTIVARIATE TIME SERIES

Fractals ◽  
2018 ◽  
Vol 26 (04) ◽  
pp. 1850058 ◽  
Author(s):  
XUEGENG MAO ◽  
PENGJIAN SHANG
Keyword(s):  
Time Series ◽  
Correlation Analysis ◽  
Cross Correlation ◽  
Coupling Parameter ◽  
Multivariate Time Series ◽  
Stock Indices ◽  
Cross Correlation Analysis ◽  
Cross Correlations ◽  
Scale Properties ◽  
Multivariate Systems

It is a crucial topic to identify the cross-correlations between time series in multivariate systems. In this paper, we extend the detrended cross-correlation analysis (DCCA) into the multivariate systems, assigned multivariate detrended cross-correlation analysis (MVDCCA). Numerical simulations of synthetic multivariate time series generated by two-exponent and mix-correlated ARFIMA processes are applied to illustrate the validity of the proposed MVDCCA. Results show that the external coupling parameter determines the strength of cross-correlation no matter that it is inter-independent or correlated among channels in a certain multivariate time series. The MVDCCA method is robust enough to detect the scale properties of time series by estimating the Hurst exponent. And we use cross-correlation coefficient to quantify the level of cross-correlations clearly. Furthermore, the MVDCCA method performs well when applied to the stock markets combining the stock daily price returns and trading volume of stock indices. By comparing results only using stock daily price returns in published literatures, we find that the higher recognizability between the pair stock indices can be observed whatever from the same regions or different regions in multivariate situations and the conclusions are more comprehensive.

Multifractal detrended cross-correlation between the Chinese domestic and international gold markets based on DCCA and DMCA methods

2014 ◽  
Vol 28 (11) ◽  
pp. 1450090 ◽  
Author(s):  
Guangxi Cao ◽  
Yan Han ◽  
Yuemeng Chen ◽  
Chunxia Yang
Keyword(s):  
Correlation Analysis ◽  
Short Range ◽  
Cross Correlation ◽  
Moving Average ◽  
Spot Markets ◽  
Artificial Data ◽  
Comparison Results ◽  
Cross Correlation Analysis ◽  
Cross Correlations ◽  
The Cross

Based on the daily price data of Shanghai and London gold spot markets, we applied detrended cross-correlation analysis (DCCA) and detrended moving average cross-correlation analysis (DMCA) methods to quantify power-law cross-correlation between domestic and international gold markets. Results show that the cross-correlations between the Chinese domestic and international gold spot markets are multifractal. Furthermore, forward DMCA and backward DMCA seems to outperform DCCA and centered DMCA for short-range gold series, which confirms the comparison results of short-range artificial data in L. Y. He and S. P. Chen [Physica A 390 (2011) 3806–3814]. Finally, we analyzed the local multifractal characteristics of the cross-correlation between Chinese domestic and international gold markets. We show that multifractal characteristics of the cross-correlation between the Chinese domestic and international gold markets are time-varying and that multifractal characteristics were strengthened by the financial crisis in 2007–2008.

The Effect of “Wuhan Closure” on the COVID-19 Pandemic in China

2021 ◽  
pp. 2150052
Author(s):  
Jian Wang ◽  
Wei Shao ◽  
Yan Yan ◽  
Junseok Kim
Keyword(s):  
Correlation Analysis ◽  
Cross Correlation ◽  
Newly Diagnosed ◽  
Daily Number ◽  
Number Of Patients ◽  
Cross Correlation Analysis ◽  
Cross Correlations ◽  
The Cross ◽  
The Impact

In this study, we analyzed daily records of newly diagnosed cases in Wuhan, Hubei excluding Wuhan (HEW), and China excluding Hubei (CEH) to investigate the impact of the new coronavirus outbreak in Wuhan on cities around it and throughout China. We used multifractal detrended cross-correlation analysis (MF-DXA) method to investigate the correlations between the daily number of patients in Wuhan and HEW as well as in Wuhan and CEH. We concluded that the cross-correlations between the daily number of patients in Wuhan and HEW were higher than those between the daily number of patients in Wuhan and CEH because the multifractal features of Wuhan and HEW are greater than those of Wuhan and CEH. We also found that the “Wuhan closure” conducted on January 23 resulted in a decrease in cross-correlations between Wuhan and CEH.

scholarly journals Segment Stock Market, Foreign Investors, and Cross-Correlation: Evidence from MF-DCCA and Spillover Index

Complexity ◽  
2020 ◽  
Vol 2020 ◽  
pp. 1-15
Author(s):  
Nan Xu ◽  
Songsong Li
Keyword(s):  
Stock Market ◽  
Stock Markets ◽  
Multifractal Analysis ◽  
Cross Correlation ◽  
Spillover Effects ◽  
The Other ◽  
Foreign Investors ◽  
Cross Correlation Analysis ◽  
Cross Correlations ◽  
The Cross

Employing the tools of multifractal detrended cross-correlation analysis (MF-DCCA) and Diebold–Yilmaz spillover index (D.Y. spillover index), we examine the effect that the foreign investors have on the cross-correlations between the two-segment stock markets, that are the accessible and the inaccessible stock markets, and the other ten respective stock markets. The shares cross-listed by the same corporates on both the A-share and H-share stock markets of China serve as the best sample to compile the two stock indices, which stands for the inaccessible stock market (AHA) and the accessible stock market (AHH), respectively. Empirical results show that the cross-correlations between the two-segment stock markets and the other ten pairs are multifractal, the multifractal strength of cross-correlations is stronger in AHH than AHA, and the intensified growth of the multifractal cross-correlations in AHA can be seen as the increasing of the openness in the inaccessible market. The empirical result of D.Y. spillover index is consistent with the multifractal analysis above, and another interesting finding is that among the selected markets, the three markets with the strongest spillover effects with AHA and AHH are Taiwan, South Korea, and Singapore, respectively, and the weakest one is Australia during the sample scenarios.

scholarly journals Joint Multifractal Analysis and Source Testing of River Level Records Based on Multifractal Detrended Cross-Correlation Analysis

Complexity ◽  
2020 ◽  
Vol 2020 ◽  
pp. 1-13
Author(s):  
Liang Wu ◽  
Manling Wang ◽  
Tongzhou Zhao
Keyword(s):  
Correlation Analysis ◽  
Multifractal Analysis ◽  
Cross Correlation ◽  
Source Analysis ◽  
Data Set ◽  
River Level ◽  
Cross Correlation Analysis ◽  
Cross Correlations ◽  
The Cross ◽  
Joint Multifractal Analysis

The joint multifractal analysis is usually conducted in two different variables for their cross-correlations but rarely used for two records of one variable collected at two different places. It is important for the detection of change in multifractality in space. Besides, the cross-correlations in two analyzed series make the analysis of sources of joint multifractality difficult. There are few studies on the source of joint multifractality. We focus on the two issues for two level records at pairs of adjacent sites along one river and carry out an extension of our previous work which is about the single multifractality of one record with the same data set. The data set is collected from 10 observation stations of a northern China river and contains about two million high-frequency river level records. Results of joint multifractal analysis via multifractal detrended cross-correlation analysis show that the change in joint multifractality at pairs of adjacent sites caused by weak cross-correlations can be detected by comparing the single generalized Hurst exponent with the joint scaling exponent function and reveal the effects of human activities on joint multifractality. This analysis provides an approach for detecting the change in multifractality. Following the idea of our previous work, two robust hypothesis tests via a set of pairs of surrogate series are proposed for the source testing of joint multifractality. The analysis of the effects of cross-correlations is carried out via a proposed simultaneously half-shifting technique which can both minimize the cross-correlations between original series and make full use of records. Results of source analysis show not only the effects of autocorrelations in series and probability distribution of river levels but also the effects of cross-correlations between series.

Sign in / Sign up

Export Citation Format

Share Document