For extremist distributions, the tail exponent (alpha) can be estimated reliably with relatively small samples using inverse gamma distribution theory because the exponent itself follows a predictable distribution, whereas estimating the mean is nearly impossible due to tail bias.