How We Tend To Overestimate Powerlaw Tail Exponents

Nassim Taleb has shared a new paper in PDF form on his Facebook Page titled: How We Tend To Overestimate Powerlaw Tail Exponents

In the presence of a layer of metaprobabilities (from metadistribution of the parameters), the asymptotic tail exponent corresponds to the lowest possible tail exponent regardless of its probability. The problem explains “Black Swan” problems, i.e., why measurements tend to chronically underestimate tail effects, rather than merely deliver imprecise but unbiased estimates.

Link: http: // www. fooled by randomness. com/ minexponents. pdf

Leave a Reply

Your email address will not be published. Required fields are marked *