No label defined (Q4935943)

This dissertation is devoted to describing, developing and (the main topic) studying comparative numerically investigation of models of financial assets based on the generalized hyperbolic (GH) distribution and its subclasses. These distributions were introduced by Barndorf-Nielsen (1977) and were applied in financial dynamics by Eberlin and Keller (1995). The author examines statistically (on the base of German stocks and other data) the fit of GH distributions to log-return distribution of assets and extendsd the hyperbolic model for the pricing of derivatives, developed by Eberlin and Keller, to GH Lévy motions. This model is compared with the classical Black-Scholes model. The author also investigates the generalized GH Lévy model where the latent volatility (mean square deviation of the growth rate of stock prices) is a stochastic process of Ornstein-Uhlenbeck type. The second part of the dissertation (Chapters 4 and 5) is devoted to problems of multivariancy of financial markets. Here the multivariate GH distribution is described and estimated on real data. The same methodology as in the first (univariate) part (but more complicated) is used for investigating such problems as calculations of prices for ``basket options'' and market risk controlling. A number of numerical results is discussed in view of computational aspects.