. "University of South Bohemia" . "It is very important that each risk model is validated, ie. it is verified whether it describes the risk on a given probability level sufficiently or not. One of the most popular methods is the backtesting, ie. utilizing of the past market data. In this paper, we focus on market risk modeling via subordinated L\u00E9vy models joined by ordinary elliptical copula functions. Selected combinations of models (geometric Brownian motion, variance gamma model, normal inverse Gaussian model for marginal distribution and Gaussian and Student copula functions for joint distribution) are assumed in order to verify the backtesting power of several combinations of normalized data as a basis for parameter estimation. It is documented that while the (linear) dependency structure is of a short memory, in order to estimate the higher moments (skewness and kurtosis) of the underlying distribution well, longer time series is required."@en . "978-80-7394-218-2" . "27510" . "RIV/61989100:27510/10:10225492!RIV11-GA0-27510___" . "Market risk backtesting via L\u00E9vy models and parameter estimation" . . "1"^^ . . "\u010Cesk\u00E9 Bud\u011Bjovice" . "[CD0BCFEE9338]" . "\u010Cesk\u00E9 Bud\u011Bjovice" . . "1"^^ . . . "Market risk backtesting via L\u00E9vy models and parameter estimation" . "Market risk backtesting via L\u00E9vy models and parameter estimation"@en . . "P(GA402/08/1237), S" . . "RIV/61989100:27510/10:10225492" . . "Tich\u00FD, Tom\u00E1\u0161" . . . "269696" . . "Backtesting, copula functions, financial institutions, market risk"@en . "2010-09-08+02:00"^^ . . . . "000287979900107" . . "It is very important that each risk model is validated, ie. it is verified whether it describes the risk on a given probability level sufficiently or not. One of the most popular methods is the backtesting, ie. utilizing of the past market data. In this paper, we focus on market risk modeling via subordinated L\u00E9vy models joined by ordinary elliptical copula functions. Selected combinations of models (geometric Brownian motion, variance gamma model, normal inverse Gaussian model for marginal distribution and Gaussian and Student copula functions for joint distribution) are assumed in order to verify the backtesting power of several combinations of normalized data as a basis for parameter estimation. It is documented that while the (linear) dependency structure is of a short memory, in order to estimate the higher moments (skewness and kurtosis) of the underlying distribution well, longer time series is required." . . "Mathematical Methods in Economics 2010" . "Market risk backtesting via L\u00E9vy models and parameter estimation"@en . "6"^^ . .