摘要：This paper uses a straightforward application of alpha-stable distributions for
Romanian Stock Market, showing how a relatively simple implementation in the real world of
a complex mathematical tool can be much more reliable in risk management than the classical
Gaussian or log-normal distributions. In this paper we use a SAS macro for estimating the
parameters of an alpha-stable distribution, using the time-series regression method from
Kogon and Williams (1998). Using the Fast Fourier Transform, we estimate the probability
density function, the cumulative distribution function and consequently, the VaR (99.5%)
and TVaR (99%). For numerical illustration we are using daily logreturns of the BET Index;
the measures of market risk, estimated on rolling windows using alpha-stable distributions
and Gaussian distribution, are then compared to the actual logreturns of the BET Index.
Numerical experiments show that using alpha-stable distributions for estimating VaR and
TVaR can be a better alternative for managing the risk of financial assets.
