ABSTRACT

Distributional assumptions on daily returns of the financial assets play a crucial role in Value-at-Risk(VaR). For years the studies have shown that the distribution of daily returns of many financial assets have heavy or semi-heavy tails. In this study, the distribution of BIST100-daily returns for the period of 2010-2016 is modelled by using Generalized Hyperbolic Distributions(GHD) which have semi-heavy tails. For this purpose, parameter estimations for GHD and their two subclasses: Normal Inverse Gaussian and Generalized Hyperbolic Skew-t Distributions are implemented and their suitabilities are tested. Finally, a VaR analysis is performed by using the estimated parameters and the performances of GHD are compared via backtesting.