Kvantifikace rizika měnových expozic s využitím Value at Risk

Abstract

The main topic of this thesis is to verify the Value at Risk at confidence level of 99 % using backtesting. The best estimate of Value at Risk is chosen through the use of backtesting. To estimate the Value at Risk are used three marginal distribution, which are normal, variance gamma and normal inverse Gaussian distribution. The work is divided into theoretical, methodological and practical part. The first part covers the characteristics of financial risk, next is the definition of Value at Risk and three methods to estimate this value. The second part is focused on selected procedures, first are describes models estimating the Value at Risk and the subsequent the statistical tests used for backtesting. The last part are models applied on yields of the historical exchange rates and compared on the basis of backtesting at interval of moving average 50, 100, 250, 500 and 1 000 days.