Modelling Value at Risk of Foreign Exchange Rates with Stable Distributions

The paper presents the theory of stable distributions with connection to financial risk. We performed an analysis of extreme values in returns of exchange rates. We compared two different approaches to compute VaR. We analyzed the returns of Exchange rates CZK, HUF and PLN, for which we find out the stable distribution parameters and test goodness of fit for these distributions. We compared the value of one-day and ten-day VaR based on normality and stable assumption. We observed that VaR under normal distribution assumption underestimate the financial risk for the higher level of confidence.