Capturing extremes

The ability of the Generalised Extreme Value and Generalised Logistic distributions to describe adequately extreme financial returns is examined. The empirical results strongly reject the Generalised Extreme Value in favour of the fatter tailed Generalised Logistic. This implies that risk measurements which are based on the Generalised Extreme Value may underestimate risk since it assigns lower probabilities to the really ruinous events located deep into the tails of the returns distribution