A note on power-law cross-correlated processes
Section snippets
Introduction: multifractality in financial data
A large amount of empirical studies have shown that stock returns exhibit fat tails and some other features that are difficult to be modeled by simple distributions. Such statistical properties are known as stylized facts and appear to be universal for all kind of financial data. In this way, Cont provides in Cont [2] an interesting review about the statistical properties of financial data.
The first stylized fact regards the fat-tailed distributions. Part of the financial literature has
Multifractal power-law cross-correlation algorithms: a literature review
The goal of all proposed methods of multifractal power-law cross-correlation analysis is to detect possible multifractal long-range cross-correlations between two time series. Applications of such techniques has been extended to different fields. The seminal paper on this topic was due to Meneveau et al., where the authors analyzed the relationship between the dissipation rates of kinetic energy and passive scalar fluctuations in fully developed turbulence [50]. Such a method is known as the
Power-law cross-correlated processes
In [62], it was highlighted that the bivariate Hurst exponent does not provide information about possible relations between a pair of series itself. Thus, a careful comparison involving the bivariate self-similarity exponent and the separate Hurst exponents of the two series becomes necessary.
The detrended cross-correlation analysis (DCCA/DXA) was introduced by Podobnik and Stanley [51] as a natural step from DFA towards power-law cross-correlation analysis. That approach is based on scaling of
Theoretical support for MFHXA
Algorithm MFHXA (see [1]) is based on the assumptions that two processes Xt and Yt satisfy the following equation (c.f. [1, Eq. (2)]):
Next, we prove that for a pair of random processes with co-affine increments of parameter H, the previous equation is satisfied and hence the MFHXA algorithm can be used to calculate the generalized bivariate Hurst exponent . Theorem 4.1 Let Xt and Yt be two random functions with co-affine increments of bivariate parameter Hx,y. Then MFHXA
Introducing MFXFD
In this section we introduce the power-law cross-correlation version of the FD algorithm to calculate the Hurst exponent.
First, recall that the FD approach (c.f. [77], [78]) is based on the following equation:
Now, we introduce the multifractal cross fractal dimension (MFXFD) method, which is valid for processes Xt and Yt that satisfies the following equation:
As always, in order to calculate Hx,y(q) for some q > 0, one must take logarithms in the
Conclusions
In this paper, we have introduced the concept of co-self-similar processes, which are the generalization of self-similar processes to the bivariate case. Moreover, we have proved that they constitute a natural context where algorithms such as the MFHXA have most sense. In addition, we have introduced a novel procedure (MFXFD) to properly calculate the bivariate parameter of such processes and empirically tested its behavior for this purpose as an alternative to DCCA or MFHXA approaches.
In
CRediT authorship contribution statement
M. Fernández-Martínez: Supervision, Conceptualization, Project administration. M.A. Sánchez-Granero: Writing - original draft, Investigation, Methodology. M.P. Casado Belmonte: Formal analysis, Validation, Software, Methodology. J.E. Trinidad Segovia: Visualization, Writing - review & editing.
Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgements
The authors would like to express their gratitude to the anonymous reviewers whose suggestions, comments, and remarks have allowed them to enhance the quality of this paper.
References (78)
Geometric stable laws: estimation and applications
Math Comput Model
(1999)- et al.
Multivariate geometric stable distributions in financial applications
Math Comput Model
(1999) - et al.
Asymmetric laplace laws and modeling financial data
Math Comput Model
(2001) - et al.
Application of multifractal measures to tehran price index
Phys A
(2005) - et al.
Multifractal detrended cross-correlation analysis between the chinese stock market and surrounding stock markets
Phys A
(2013) - et al.
Multifractality and herding behavior in the japanese stock market
Chaos Solitons Fractals
(2009) - et al.
Unifractality and multifractality in the italian stock market
Int Rev Financ Anal
(2009) - Kukacka J., Kristoufek L. Do “complex” financial models really lead to complex dynamics? Agent-based models and...
Alternative measure of multifractal content and its application in finance
Chaos Solitons Fractals
(2016)- et al.
Quantitative approach to multifractality induced by correlations and broad distribution of data
Phys A
(2018)
Singularity spectrum of fractal signals from wavelet analysis: exact results
J Stat Phys
The Markov-switching multifractal model of asset returns: GMM estimation and linear forecasting of volatility
J Bus Econ Stat
Understanding volatility correlation behavior with a magnitude cross-correlation function
Phys Rev E
Cross-correlation detection and analysis for California’s electricity market based on analogous multifractal analysis
Chaos
Multifractal cross wavelet analysis
Fractals
Multifractal temporally weighted detrended cross-correlation analysis to quantify power-law cross-correlation and its application to stock markets
Chaos
Detrended cross-correlation analysis consistently extended to multifractality
Phys Rev E
Can the bivariate hurst exponent be higher than an average of the separate hurst exponents?
Phys A
Power-law cross-correlations estimation under heavy tails
Commun Nonlinear Sci Numer Simul
On the interplay between short and long term memory in the power-law cross-correlations setting
Phys A
Spectrum-based estimators of the bivariate hurst exponent
Phys Rev E
DCCA cross-correlation coefficient: quantifying level of cross-correlation
Phys A
Detrending moving-average cross-correlation coefficient: measuring cross-correlations between non-stationary series
Phys A
Detrended fluctuation analysis as a regression framework: estimating dependence at different scales
Phys Rev E
Multifractal height cross-correlation analysis: a new method for analyzing long-range cross-correlations
EPL (Europhys Lett)
Empirical properties of asset returns: stylized facts and statistical issues
Quant Finance
The variation of certain speculative prices
J Bus
Financial power laws: empirical evidence, models, and mechanisms
Chaos Solitons Fractals
The fine structure of asset returns: an empirical investigation
J Bus
The modified tempered stable distribution, GARCH models and option pricing
Probab Math Stat
The limiting distribution of extremal exchange rate returns
J Appl Econom
Scaling in stock market data: stable laws and beyond
Inverse cubic law for the distribution of stock price variations
Eur Phys J B
The oxford handbook of computational economics and finance, chapter multifractal models in finance: their origin, properties, and applications
A multifractal walk down wall street
Sci Am
Multifractality in asset returns: theory and evidence
Rev Econ Stat
Sparseness and roughness of foreign exchange rates
Int J Modern Phys C
Multifractal analysis of foreign exchange data
Appl Stoch Models Data Anal
Cited by (2)
Extending DFA-based multiple linear regression inference: Application to acoustic impedance models
2021, Physica A: Statistical Mechanics and its ApplicationsCitation Excerpt :More precisely, a variable with a wider scale could produce larger regression coefficients, without actually being more important. On the other hand, since power-law cross-correlations are not equal to correlations [25], multicollinearity issues cannot be excluded, or more precisely, it is not possible to ascertain yet if any other sources of bias come with partial cross-correlations. Therefore, some other measure of variable importance, taking into account the scales magnitude may be more appropriate.
POWER-LAW LÉVY PROCESSES, POWER-LAW VECTOR RANDOM FIELDS, AND SOME EXTENSIONS
2023, Proceedings of the American Mathematical Society
- 1
Partially supported by Ministerio de Ciencia, Innovación y Universidades, Spain and FEDER, Spain, grant PGC2018-097198-B-I00.
- 2
Partially supported by Fundación Séneca of Región de Murcia (Murcia, Spain), grant 20783/PI/18.
- 3
Partially supported by Ministerio de Ciencia, Innovación y Universidades, Spain and FEDER, Spain, grant PGC2018-101555-B-I00.
- 4
Partially supported by UAL/CECEU/FEDER, Spain, grant UAL18-FQM-B038-A.
- 5
Author acknowledges the support of CDTIME.