"978-80-86175-87-4" . . "[6662967331C6]" . "93671" . "000339103100124" . "Volatility can be defined and measured as a risk of financial instrument over a specified time period. In this paper, we deal with volatility model selection and comparison in a specific framework. In particular, univariate volatility models like traditional ARCH model and its extensions will be object of our interest. Selection of the best suitable model may be usually based on in-sample or out-of-sample criteria. In empirical studies, we usually favour model that can capture real features of the data analysed and, in addition, can provide the most accurate out-of-sample forecast quality. In this paper, we focus just on out-of-sample comparison of linear and nonlinear ARCH family models which may follow two different approaches. In first approach, alternative models are contrasted by different loss functions based directly on variance forecast and Diebold-Mariano type tests. The second approach includes indirect evaluation methods which conside r using of alternative variance forecasts. In our study, we deal with the evaluation of alternative ARCH family models within a VaR framework. Empirical comparison of the methods discussed above will be demonstrated on illustrative example using sample data from U.S. stock market. We consider daily data of S&P500 index in the period of 2007-2012 years which includes the stage of recent global financial crisis of 2008-2009 years."@en . "The 7th International Days of Statistics and Economics : conference proceedings : September 19-21, 2013, Prague, Czech Republic" . . "27510" . . "Volatility can be defined and measured as a risk of financial instrument over a specified time period. In this paper, we deal with volatility model selection and comparison in a specific framework. In particular, univariate volatility models like traditional ARCH model and its extensions will be object of our interest. Selection of the best suitable model may be usually based on in-sample or out-of-sample criteria. In empirical studies, we usually favour model that can capture real features of the data analysed and, in addition, can provide the most accurate out-of-sample forecast quality. In this paper, we focus just on out-of-sample comparison of linear and nonlinear ARCH family models which may follow two different approaches. In first approach, alternative models are contrasted by different loss functions based directly on variance forecast and Diebold-Mariano type tests. The second approach includes indirect evaluation methods which conside r using of alternative variance forecasts. In our study, we deal with the evaluation of alternative ARCH family models within a VaR framework. Empirical comparison of the methods discussed above will be demonstrated on illustrative example using sample data from U.S. stock market. We consider daily data of S&P500 index in the period of 2007-2012 years which includes the stage of recent global financial crisis of 2008-2009 years." . . . "Praha" . . "RIV/61989100:27510/13:86086782!RIV15-MSM-27510___" . . "Slan\u00FD" . . . . . "RIV/61989100:27510/13:86086782" . "On Conditional Volatility Model Comparison"@en . . . "On Conditional Volatility Model Comparison"@en . . "Se\u010Fa, Petr" . "conditional volatility, Diebold-Mariano test, MCS approach, out-of-sample forecast, VaR"@en . "On Conditional Volatility Model Comparison" . "On Conditional Volatility Model Comparison" . . "P(EE2.3.20.0296), P(GA13-13142S)" . . "1"^^ . "10"^^ . "Melandrium" . . . . "2013-09-19+02:00"^^ . "1"^^ .