Joanna Gorka, Ph.D., Department of Econometrics and Statistics, Nicolaus Copernicus University of Toruń, ul. Gagarina 13 A, Torun, 87-100, Poland
Value-at-Risk (VaR) is used as a tool for measuring the market risk. Alternative VaR implementation has been but none of known methods are not the best. This paper proposes to use a family sign RCA models (Thavaneswaran et al. 2006) to estimate Value-at-Risk for selected shares from the Warsaw Stock Exchange. Obtained VaR forecasts out of Sign RCA models and ARMA-(G)ARCH models will compare. The forecasts will evaluate using traditional tests and loss function method. We expect sign RCA models will be dominate over other models of VaR forecasting for some share.