arxiv: 2605.10303 · v2 · submitted 2026-05-11 · 🧮 math.ST · stat.TH

Recognition: 2 theorem links

· Lean Theorem

Measuring Tail Dependence in Linear Processes: Theory and Empirics

Debanjana Datta , Diganta Mukherjee

Authors on Pith no claims yet

Pith reviewed 2026-05-14 21:53 UTC · model grok-4.3

classification 🧮 math.ST stat.TH

keywords tail dependencelinear processesregular variationextreme co-movementspersistencedependence measureheavy tailscryptocurrency data

0 comments

The pith

A dependence measure quantifies tail dependence in linear processes with regularly varying distributions, including persistence effects and both identical and non-identical cases.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper introduces a dependence measure to describe joint extremes in time series that have heavy tails and extreme co-movements, features that Gaussian models miss. It focuses on linear processes whose marginals are regularly varying and shows how the measure incorporates persistence without needing exact tail indices or specific innovation distributions. The approach is tested on high-frequency cryptocurrency data to study persistence effects and is checked through simulations that confirm the theoretical expectations.

Core claim

The central claim is that a new dependence measure for tail dependence in linear processes works for both identical and non-identical regularly varying distributions. This measure captures the effect of persistence on extreme co-movements. Theoretical results, analysis of cryptocurrency datasets, and simulation studies demonstrate that the measure detects these persistence effects without further specification of the tail indices or innovation distributions.

What carries the argument

The tail dependence measure defined on linear processes with regularly varying marginals that incorporates persistence effects.

If this is right

The measure applies to analysis of extreme co-movements in financial series that exhibit persistence.
It extends to cases with non-identical regularly varying distributions.
Simulation results show the measure accurately reflects persistence in tail dependence.
Empirical cryptocurrency data illustrate how persistence influences joint extremes.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

The measure could be used to compare tail dependence across different asset classes with heavy tails.
Extensions might examine how the measure behaves in processes that are nearly linear but contain small nonlinear terms.
It could inform risk models that need to account for clustered extremes in volatile markets.

Load-bearing premise

The time series are linear processes whose marginal distributions are regularly varying.

What would settle it

A simulation of a linear process with regularly varying tails where the measure shows no change in dependence when persistence parameters are varied would falsify the claim that it captures persistence effects.

read the original abstract

The quantitative analysis of financial time series often reveals two distinct features that standard Gaussian frameworks fail to capture: heavy-tailed marginal distributions and the phenomenon of extreme co-movements.While extreme value theory characterizes marginal behavior, Copulas provide a functional bridge to describe the dependence structure independently of the marginals. We are proposing a different way of looking at the joint extremes on the basis of a dependence measure. The proposed idea incorporates both the non-identical and identical regularly varying distributions. Informed by the analysis of some high-frequency cryptocurrency datasets, the effect of persistence property have been thoroughly studied under these setups. A detailed simulation study confirms our intuition and findings.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

The paper gives a dependence measure for tail co-movements in linear processes that covers both identical and non-identical regularly varying marginals, backed by simulations and crypto data, but the non-identical case needs a close check on whether extra coefficient restrictions are hidden.

read the letter

The main point is a new dependence measure for joint extremes in linear processes that is meant to work whether the marginal tails are identical or not. It builds on regularly varying theory and tries to give a direct way to quantify persistence in extremes without going through copulas first. That is the explicit extension they highlight. The simulations confirm the basic behavior and the high-frequency cryptocurrency application shows how persistence looks in real data, which adds some practical grounding for risk work. Those parts are straightforward and useful for readers who already work in this area. The soft spot is the technical handling of non-identical tails. When the tail indices differ, the joint regular variation of the vector process usually requires conditions on the linear coefficients so the limit measure does not collapse to the axes. The abstract claims the construction works generally, but without the explicit formulas or error bounds it is hard to see whether those conditions are stated or avoided. The stress-test concern lands here until the derivations are checked. This is for people already doing extreme-value work on financial time series. A reader who needs a tool for heterogeneous tails and persistence would get value from the proposal and the data example. It deserves peer review so the math can be verified against the non-identical case.

Referee Report

2 major / 1 minor

Summary. The manuscript proposes a novel dependence measure for capturing joint extremes (tail dependence) in linear processes with regularly varying marginals. The measure is designed to handle both identical and non-identical tail indices, and the authors study persistence effects using high-frequency cryptocurrency data, with a simulation study to confirm the theoretical findings.

Significance. If the central construction is shown to be rigorous and free of hidden restrictions when marginal tail indices differ, the work would provide a useful alternative perspective to copula-based methods for analyzing extreme co-movements in heavy-tailed time series, with direct relevance to financial applications such as cryptocurrency risk modeling.

major comments (2)

[Abstract / Theoretical development] Abstract and theoretical section: the claim that the dependence measure works for non-identical regularly varying marginals (α_i ≠ α_j) in general linear processes X_t = sum A_j Z_{t-j} is not accompanied by explicit conditions on the coefficient matrices A_j or the spectral measures that guarantee the joint regular variation limit exists in the required form; without these, the vague convergence of the normalized point process may concentrate on the axes or fail to exist, undermining the generality asserted for the persistence analysis.
[Simulation study] Simulation study: the abstract states that simulations confirm the findings, yet no details are provided on how the measure is computed when tail indices differ, nor on error bounds or sensitivity to innovation distributions; this makes it impossible to verify whether the reported persistence effects are robust or influenced by post-hoc choices.

minor comments (1)

[Abstract] The abstract contains a long run-on sentence beginning 'While extreme value theory...'; splitting it would improve readability.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive comments, which highlight important points for strengthening the theoretical foundations and simulation details. We address each major comment below.

read point-by-point responses

Referee: [Abstract / Theoretical development] Abstract and theoretical section: the claim that the dependence measure works for non-identical regularly varying marginals (α_i ≠ α_j) in general linear processes X_t = sum A_j Z_{t-j} is not accompanied by explicit conditions on the coefficient matrices A_j or the spectral measures that guarantee the joint regular variation limit exists in the required form; without these, the vague convergence of the normalized point process may concentrate on the axes or fail to exist, undermining the generality asserted for the persistence analysis.

Authors: We agree that the current presentation lacks sufficient explicit conditions for the non-identical case. In the revised manuscript we will add Assumption 2.3, requiring that the coefficient matrices satisfy sum_j ||A_j||^β < ∞ for all β < min(α_i, α_j) and that the spectral measures Γ_i and Γ_j are mutually absolutely continuous with respect to Lebesgue measure on the positive orthant so that the limiting intensity measure charges the interior of the first quadrant. A short proof that these conditions prevent concentration on the axes will be included in the appendix, and the abstract will be updated to reference the strengthened assumptions. revision: yes
Referee: [Simulation study] Simulation study: the abstract states that simulations confirm the findings, yet no details are provided on how the measure is computed when tail indices differ, nor on error bounds or sensitivity to innovation distributions; this makes it impossible to verify whether the reported persistence effects are robust or influenced by post-hoc choices.

Authors: We accept that the simulation section is underspecified. The revised version will expand Section 4 with: (i) explicit pseudocode showing separate normalization by the estimated tail indices when α_i ≠ α_j; (ii) Monte Carlo standard errors obtained from 5000 replications; and (iii) a new sensitivity table comparing results under Student-t, Pareto, and stable innovations. The persistence patterns remain qualitatively unchanged across these distributions. revision: yes

Circularity Check

0 steps flagged

No circularity: tail dependence measure defined directly from regular variation and linear process structure

full rationale

The paper defines its dependence measure for joint extremes in linear processes from the regularly varying marginals (identical or non-identical) and the linear filter representation, without any reduction of the measure to a fitted parameter, self-referential definition, or load-bearing self-citation. The abstract and described setup present the construction as following from the vague convergence properties of the normalized point process under the stated assumptions, with simulations and crypto data serving as validation rather than input. No equations or steps in the provided text exhibit the patterns of self-definition, fitted-input prediction, or ansatz smuggling via citation; the derivation chain remains independent of its target outputs.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The central claim rests on the domain assumption that the time series follow linear processes with regularly varying marginals; no free parameters or invented entities are explicitly introduced in the abstract.

axioms (1)

domain assumption The time series are linear processes with regularly varying distributions (identical or non-identical).
Invoked to define the tail dependence measure and study persistence effects.

pith-pipeline@v0.9.0 · 5397 in / 1186 out tokens · 29444 ms · 2026-05-14T21:53:05.329978+00:00 · methodology

Review history (2 revisions) →

discussion (0)

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

IndisputableMonolith/Cost/FunctionalEquation.lean washburn_uniqueness_aczel unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

The tail cross-correlation coefficient at lag k is defined as τ_{x_n,y_n}(k) = P(Y^*_{t+k} > y_n | X^*_t > x_n) − P(Y^*_{t+k} > y_n) / (1 − P(X^*_t > x_n))
IndisputableMonolith/Foundation/RealityFromDistinction.lean reality_from_one_distinction unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

Theorem 1 (Samorodnitsky 2016) on balanced regularly varying tails of linear combinations

What do these tags mean?

matches: The paper's claim is directly supported by a theorem in the formal canon.
supports: The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends: The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses: The paper appears to rely on the theorem as machinery.
contradicts: The paper's claim conflicts with a theorem or certificate in the canon.
unclear: Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

30 extracted references · 30 canonical work pages · 1 internal anchor

[1]

Biometrika , volume=

High-quantile regression for tail-dependent time series , author=. Biometrika , volume=. 2021 , publisher=

work page 2021
[2]

North American Actuarial Journal , volume=

On an approximation for the surplus process using extreme value theory: Applications in ruin theory and reinsurance pricing , author=. North American Actuarial Journal , volume=. 2004 , publisher=

work page 2004
[3]

ASTIN Bulletin: The Journal of the IAA , volume=

An extreme-value theory approximation scheme in reinsurance and insurance-linked securities , author=. ASTIN Bulletin: The Journal of the IAA , volume=. 2018 , publisher=

work page 2018
[4]

ISOPE International Ocean and Polar Engineering Conference , pages=

Joint probability analysis of hurricane Katrina 2005 , author=. ISOPE International Ocean and Polar Engineering Conference , pages=. 2006 , organization=

work page 2005
[5]

bmw williamsf1 team , pages=

Applications of extreme value theory to collateral valuation , author=. bmw williamsf1 team , pages=

work page
[6]

Economic Theory , pages=

Debt collateralization, capital structure, and maximal leverage , author=. Economic Theory , pages=. 2020 , publisher=

work page 2020
[7]

Communications in Statistics: Case Studies, Data Analysis and Applications , volume=

Analysis of hurricane extremes and record values in the Atlantic , author=. Communications in Statistics: Case Studies, Data Analysis and Applications , volume=. 2019 , publisher=

work page 2019
[8]

ASTIN Bulletin: The Journal of the IAA , volume=

On the use of extreme values to estimate the premium for an excess of loss reinsurance , author=. ASTIN Bulletin: The Journal of the IAA , volume=. 1964 , publisher=

work page 1964
[9]

Biometrika , pages=

A nonparametric estimation procedure for bivariate extreme value copulas , author=. Biometrika , pages=. 1997 , publisher=

work page 1997
[10]

The annals of Statistics , pages=

The dip test of unimodality , author=. The annals of Statistics , pages=. 1985 , publisher=

work page 1985
[11]

International statistical review , volume=

The t copula and related copulas , author=. International statistical review , volume=. 2005 , publisher=

work page 2005
[12]

Proceedings of the national Academy of Sciences , volume=

A central limit theorem and a strong mixing condition , author=. Proceedings of the national Academy of Sciences , volume=

work page
[13]

Proceedings of the National Academy of Sciences , volume=

Nonlinear system theory: Another look at dependence , author=. Proceedings of the National Academy of Sciences , volume=. 2005 , publisher=

work page 2005
[14]

Available at SSRN 317122 , year=

Beyond correlation: Extreme co-movements between financial assets , author=. Available at SSRN 317122 , year=

work page
[15]

Statistical Tools for Finance and Insurance , editor=

Jajuga, Krzysztof and Papla, Daniel , title=. Statistical Tools for Finance and Insurance , editor=. 2005 , publisher=

work page 2005
[16]

The Annals of Applied Probability , volume=

The supremum of a negative drift random walk with dependent heavy-tailed steps , author=. The Annals of Applied Probability , volume=. 2000 , publisher=

work page 2000
[17]

The Annals of Applied Probability , volume=

Tail index estimation for dependent data , author=. The Annals of Applied Probability , volume=. 1998 , publisher=

work page 1998
[18]

Stochastic processes and their applications , volume=

Tails of subordinated laws: the regularly varying case , author=. Stochastic processes and their applications , volume=. 1996 , publisher=

work page 1996
[19]

Geddes, K. O. and Czapor, S. R. and Labahn, G. Algorithms for C omputer A lgebra. 1992

work page 1992
[20]

Software engineering---from auxiliary to key technologies

Broy, M. Software engineering---from auxiliary to key technologies. Software Pioneers. 1992

work page 1992
[21]

Conductive P olymers. 1981

work page 1981
[22]

Smith, S. E. Neuromuscular blocking drugs in man. Neuromuscular junction. H andbook of experimental pharmacology. 1976

work page 1976
[23]

Chung, S. T. and Morris, R. L. Isolation and characterization of plasmid deoxyribonucleic acid from Streptomyces fradiae. 1978

work page 1978
[24]

and AghaKouchak, A

Hao, Z. and AghaKouchak, A. and Nakhjiri, N. and Farahmand, A. Global integrated drought monitoring and prediction system (GIDMaPS) data sets. 2014

work page 2014
[25]

Babichev, S. A. and Ries, J. and Lvovsky, A. I. Quantum scissors: teleportation of single-mode optical states by means of a nonlocal single photon. 2002

work page 2002
[26]

Wormholes in Maximal Supergravity

Beneke, M. and Buchalla, G. and Dunietz, I. Mixing induced CP asymmetries in inclusive B decays. Phys. L ett. 1997. arXiv:0707.3168

work page internal anchor Pith review Pith/arXiv arXiv 1997
[27]

deep SIP : deep learning of S upernova I a P arameters

Stahl, B. deep SIP : deep learning of S upernova I a P arameters. 2020. ascl:2006.023

work page 2020
[28]

Abbott, T. M. C. and others. Dark Energy Survey Year 1 Results: Constraints on Extended Cosmological Models from Galaxy Clustering and Weak Lensing. Phys. Rev. D. 2019. doi:10.1103/PhysRevD.99.123505. arXiv:1810.02499

work page doi:10.1103/physrevd.99.123505 2019
[29]

2013 , address=

Modelling extremal events: for insurance and finance , author=. 2013 , address=

work page 2013
[30]

2016 , publisher=

Stochastic Processes and Long Range Dependence , author=. 2016 , publisher=

work page 2016