The paper introduces a novel approach applying extreme value theory to fat-tailed variables that have a remote but finite upper bound.

True: the paper presents a method for applying extreme value theory to fat-tailed variables with a finite but remote upper bound (for example a log-transformation or 'dual distribution'), and no earlier published work introduced substantially the same approach. Mostly true: the paper does present such a method, but closely related earlier work exists, so its novelty is somewhat overstated. Mixed: the method is mainly an adaptation of existing techniques, with only modest new elements. Mostly false: the paper applies extreme value theory to bounded fat-tailed data, but essentially the same approach had already been published. False: the paper contains no such method.

Set before the research began.

1. E1 Extreme value statistics for truncated Pareto-type distributions arXiv (ar5iv rendering)

   “This approach appears to be new and allows to bridge tail models with \(\xi=-1\) and Pareto models in the sense that the Pareto-type distributions are considered as limits of truncated Pareto-type distributions.”

2. E2 Extreme value statistics for truncated Pareto-type distributions arXiv (ar5iv rendering)

   “Recently attention has been drawn to practical problems with the use of unbounded Pareto distributions, for instance when there are natural upper bounds that truncate the probability tail.”

3. E3 Parameter Estimation for the Truncated Pareto Distribution Journal of the American Statistical Association (Aban, Meerschaert, Panorska) · March 2006

   “InsomeapplicationsthereisanaturalupperboundontheprobabilitytailthattruncatestheParetolaw.”

4. E4 Parameter Estimation for the Truncated Pareto Distribution Journal of the American Statistical Association (Aban, Meerschaert, Panorska) · March 2006

   “Alloftheresultsinthissectionandtheprevioussectionalsoapplytothatcase,becauseY=lnXhasatruncatedshiftedexponentialdistributionwithP(Y>y)=e−α(y−lnγ)−(γ/ν)α1−(γ/ν)αforlnγ≤y≤lnνifandonlyifXhasatruncatedParetodistribution.”

None of the evidence describes the content of the subject's 2015 paper, so nothing independent shows what method it presents or how it compares with earlier work. E3 and E4 (JASA, 2006) show that earlier work estimated parameters for truncated Pareto distributions, including through a log transformation. E1 and E2 show that other authors have also worked on extreme value statistics for truncated Pareto-type distributions, but the material gives no date for that work. This suggests related earlier work existed, but without a description of the paper's own approach, the evidence cannot show whether it is substantially the same, a modest adaptation or new.