scholarly journals Heavy-tailed distributions, correlations, kurtosis and Taylor’s Law of fluctuation scaling

2020 ◽  
Vol 476 (2244) ◽  
pp. 20200610
Author(s):  
Joel E. Cohen ◽  
Richard A. Davis ◽  
Gennady Samorodnitsky
Keyword(s):  
Power Law ◽  
Statistical Models ◽  
Natural Conditions ◽  
Heavy Tailed Distributions ◽  
Sample Correlation ◽  
Taylor’S Law ◽  
Heavy Tailed ◽  
Fluctuation Scaling ◽  
Random Limit ◽  
Asymptotic Scaling

Pillai & Meng (Pillai & Meng 2016 Ann. Stat. 44 , 2089–2097; p. 2091) speculated that ‘the dependence among [random variables, rvs] can be overwhelmed by the heaviness of their marginal tails ·· ·’. We give examples of statistical models that support this speculation. While under natural conditions the sample correlation of regularly varying (RV) rvs converges to a generally random limit, this limit is zero when the rvs are the reciprocals of powers greater than one of arbitrarily (but imperfectly) positively or negatively correlated normals. Surprisingly, the sample correlation of these RV rvs multiplied by the sample size has a limiting distribution on the negative half-line. We show that the asymptotic scaling of Taylor’s Law (a power-law variance function) for RV rvs is, up to a constant, the same for independent and identically distributed observations as for reciprocals of powers greater than one of arbitrarily (but imperfectly) positively correlated normals, whether those powers are the same or different. The correlations and heterogeneity do not affect the asymptotic scaling. We analyse the sample kurtosis of heavy-tailed data similarly. We show that the least-squares estimator of the slope in a linear model with heavy-tailed predictor and noise unexpectedly converges much faster than when they have finite variances.

Taylor’s law of fluctuation scaling for semivariances and higher moments of heavy-tailed data

2021 ◽  
Vol 118 (46) ◽  
pp. e2108031118
Author(s):  
Mark Brown ◽  
Joel E. Cohen ◽  
Chuan-Fa Tang ◽  
Sheung Chi Phillip Yam
Keyword(s):  
Scaling Laws ◽  
Higher Moments ◽  
Direct Proportion ◽  
Sample Mean ◽  
Scaling Relationships ◽  
Heavy Tailed Distributions ◽  
Taylor’S Law ◽  
Heavy Tailed ◽  
Fluctuation Scaling ◽  
Sortino Ratio

We generalize Taylor’s law for the variance of light-tailed distributions to many sample statistics of heavy-tailed distributions with tail index α in (0, 1), which have infinite mean. We show that, as the sample size increases, the sample upper and lower semivariances, the sample higher moments, the skewness, and the kurtosis of a random sample from such a law increase asymptotically in direct proportion to a power of the sample mean. Specifically, the lower sample semivariance asymptotically scales in proportion to the sample mean raised to the power 2, while the upper sample semivariance asymptotically scales in proportion to the sample mean raised to the power (2−α)/(1−α)>2. The local upper sample semivariance (counting only observations that exceed the sample mean) asymptotically scales in proportion to the sample mean raised to the power (2−α2)/(1−α). These and additional scaling laws characterize the asymptotic behavior of commonly used measures of the risk-adjusted performance of investments, such as the Sortino ratio, the Sharpe ratio, the Omega index, the upside potential ratio, and the Farinelli–Tibiletti ratio, when returns follow a heavy-tailed nonnegative distribution. Such power-law scaling relationships are known in ecology as Taylor’s law and in physics as fluctuation scaling. We find the asymptotic distribution and moments of the number of observations exceeding the sample mean. We propose estimators of α based on these scaling laws and the number of observations exceeding the sample mean and compare these estimators with some prior estimators of α.

From the central limit theorem to heavy-tailed distributions

2003 ◽  
Vol 40 (3) ◽  
pp. 803-806 ◽  
Author(s):  
Jinwen Chen
Keyword(s):  
Central Limit Theorem ◽  
Limit Theorem ◽  
Central Limit ◽  
Power Law ◽  
Power Law Distribution ◽  
The Central Limit Theorem ◽  
Heavy Tailed Distributions ◽  
Heavy Tailed ◽  
Random Indices ◽  
Power Law Distributions

It has been observed that in many practical situations randomly stopped products of random variables have power law distributions. In this note we show that, in order for such a product to have a power law distribution, the only random indices are the exponentially distributed ones. We also consider a more general problem, which is closely related to problems concerning transformation from the central limit theorem to heavy-tailed distributions.

Heavy-tailed distributions for building stock data

2018 ◽  
Vol 46 (7) ◽  
pp. 1281-1296 ◽  
Author(s):  
Patrick Erik Bradley ◽  
Martin Behnisch
Keyword(s):  
Power Law ◽  
Building Stock ◽  
Underlying Distribution ◽  
Large Cities ◽  
Statistical Framework ◽  
Heavy Tailed Distributions ◽  
Heavy Tailed Distribution ◽  
Event Histories ◽  
Heavy Tailed ◽  
Log Normal

The question of inferring the owner of a set of building stocks (e.g. from which country the buildings are taken) from building-related quantities like number of buildings or types of building event histories necessitates the knowledge of their distributions in order to compare them. If the distribution function is a power law, then a version of the 80/20 rule can be applied to describe the variable. This distribution is an example of a heavy-tailed distribution; another example is the log-normal distribution. Heavy-tailed distributions have the property that studying the effects of the few large values already yields most of the overall effect of the whole quantity. For example, if reducing the CO2 emissions of the buildings of a country is the issue, then in case of a heavy-tailed distribution, only the effects of the relatively few large cities need to be considered. It is shown that the number of buildings in German municipalities or counties or the number of building-related event histories of a certain vanished building stock follow a heavy-tailed distribution and give evidence for the type of underlying distribution. The methodology used is a recent statistical framework for discerning power law and other heavy-tailed distributions in empirical data.

scholarly journals Sample correlations of infinite variance time series models: an empirical and theoretical study

1998 ◽  
Vol 11 (3) ◽  
pp. 255-282 ◽  
Author(s):  
Jason Cohen ◽  
Sidney Resnick ◽  
Gennady Samorodnitsky
Keyword(s):  
Time Series ◽  
Correlation Function ◽  
Random Permutation ◽  
Infinite Variance ◽  
Finite Variance ◽  
More Care ◽  
Sample Correlation ◽  
Heavy Tailed ◽  
Random Limit ◽  
Permutation Scheme

When the elements of a stationary ergodic time series have finite variance the sample correlation function converges (with probability 1) to the theoretical correlation function. What happens in the case where the variance is infinite? In certain cases, the sample correlation function converges in probability to a constant, but not always. If within a class of heavy tailed time series the sample correlation functions do not converge to a constant, then more care must be taken in making inferences and in model selection on the basis of sample autocorrelations. We experimented with simulating various heavy tailed stationary sequences in an attempt to understand what causes the sample correlation function to converge or not to converge to a constant. In two new cases, namely the sum of two independent moving averages and a random permutation scheme, we are able to provide theoretical explanations for a random limit of the sample autocorrelation function as the sample grows.

An analysis of power law distributions and tipping points during the global financial crisis

2018 ◽  
Vol 13 (1) ◽  
pp. 80-91 ◽  
Author(s):  
Yifei Li ◽  
Lei Shi ◽  
Neil Allan ◽  
John Evans
Keyword(s):  
Financial Crisis ◽  
Power Law ◽  
Global Financial Crisis ◽  
Operational Risk ◽  
Tipping Point ◽  
Power Law Distribution ◽  
Heavy Tailed Distributions ◽  
Heavy Tailed ◽  
The Global Financial Crisis ◽  
Power Law Distributions

AbstractHeavy-tailed distributions have been observed for various financial risks and papers have observed that these heavy-tailed distributions are power law distributions. The breakdown of a power law distribution is also seen as an indicator of a tipping point being reached and a system then moves from stability through instability to a new equilibrium. In this paper, we analyse the distribution of operational risk losses in US banks, credit defaults in US corporates and market risk events in the US during the global financial crisis (GFC). We conclude that market risk and credit risk do not follow a power law distribution, and even though operational risk follows a power law distribution, there is a better distribution fit for operational risk. We also conclude that whilst there is evidence that credit defaults and market risks did reach a tipping point, operational risk losses did not. We conclude that the government intervention in the banking system during the GFC was a possible cause of banks avoiding a tipping point.

Taylor’s law and heavy-tailed distributions

2021 ◽  
Vol 118 (50) ◽  
pp. e2118893118
Author(s):  
W. Brent Lindquist ◽  
Svetlozar T. Rachev
Keyword(s):  
Heavy Tailed Distributions ◽  
Taylor’S Law ◽  
Heavy Tailed

From the central limit theorem to heavy-tailed distributions

2003 ◽  
Vol 40 (03) ◽  
pp. 803-806
Author(s):  
Jinwen Chen
Keyword(s):  
Central Limit Theorem ◽  
Limit Theorem ◽  
Central Limit ◽  
Power Law ◽  
Power Law Distribution ◽  
The Central Limit Theorem ◽  
Heavy Tailed Distributions ◽  
Heavy Tailed ◽  
Random Indices ◽  
Power Law Distributions

It has been observed that in many practical situations randomly stopped products of random variables have power law distributions. In this note we show that, in order for such a product to have a power law distribution, the only random indices are the exponentially distributed ones. We also consider a more general problem, which is closely related to problems concerning transformation from the central limit theorem to heavy-tailed distributions.

scholarly journals Large-Scale, Wavelet-Based Analysis of Lysosomal Trajectories and Co-Movements of Lysosomes with Nanoparticle Cargos

Cells ◽  
2022 ◽  
Vol 11 (2) ◽  
pp. 270
Author(s):  
Konstantin Polev ◽  
Diana V. Kolygina ◽  
Kristiana Kandere-Grzybowska ◽  
Bartosz A. Grzybowski
Keyword(s):  
Power Law ◽  
Large Scale ◽  
Cell Types ◽  
Potential Models ◽  
Movement Trajectories ◽  
Heavy Tailed Distributions ◽  
Nanoparticle Aggregates ◽  
Heavy Tailed ◽  
Acidic Organelles ◽  
Cancerous Cells

Lysosomes—that is, acidic organelles known for degradation/recycling—move through the cytoplasm alternating between bursts of active transport and short, diffusive motions or even pauses. While their mobility is essential for lysosomes’ fusogenic and non-fusogenic interactions with target organelles, their movements have not been characterized in adequate detail. Here, large-scale statistical analysis of lysosomal movement trajectories reveals that lysosome trajectories in all examined cell types—both cancer and noncancerous ones—are superdiffusive and characterized by heavy-tailed distributions of run and flight lengths. Consideration of Akaike weights for various potential models (lognormal, power law, truncated power law, stretched exponential, and exponential) indicates that the experimental data are best described by the lognormal distribution, which, in turn, can be related to one of the space-search strategies particularly effective when “thorough” search needs to balance search for rare target(s) (organelles). In addition, automated, wavelet-based analysis allows for co-tracking the motions of lysosomes and the cargos they carry—particularly the nanoparticle aggregates known to cause selective lysosome disruption in cancerous cells. The methods we describe here could help study nanoparticle assemblies, viruses, and other objects transported inside various vesicle types, as well as coordinated movements of organelles/particles in the cytoplasm. Custom-written code that includes integrated workflow for our analyses is made available for academic use.

Sign in / Sign up

Export Citation Format

Share Document