StatisticalConsequencesofFatTails

This book investigates how conventional statistical methods can be misused when attempting to analyze non-standard distributions, and suggests solutions whenever possible. Simply changing the methods used to analyze data does not always work when dealing with distributions with fat tails. Traditional statistical theory deals mostly with very small or very large numbers, but the real world is somewhere in the middle. This means that in order to properly analyze non-standard distributions, researchers need to understand how the laws of the medium numbers work for each specific case. Even well-understood statistical concepts like the law of large numbers and the generalized central limit theorem do not always work outside of the standard Gaussian or Levy-Stable distributions. Some common problems include sample means not aligning with population means, difficulty in estimating empirical distributions, compounding uncertainty in statistical metrics, failure of dimension reduction techniques, biases in psychology research becoming rational under new probability distributions, and mistakes in economic and behavioral analysis stemming from using the wrong distributions. This first installment of the Technical Incerto series frames these issues in a narrative built around previously published journals articles.

作者简介

NassimNicholasTalebspent20yearsasaderivativesandmathematicaltraderbeforestartinghissecondcareerinappliedprobability.Heistheauthorof5-volumeIncerto,anessayonuncertainty,publishedin40languages–withparalleljournalarticlesandtechnicalcommentariesofwhichthisbookisanorganizedcompilation.TalebiscurrentlyDistinguishedProfessorofRis...