*
* Example 7.4 from page 299
*
open data d-intc7297.dat
data(format=free,org=columns) 1 6329 intel
*
* This produces slightly different results than shown because RATS determines the
* quantiles by taking 1+p(n-1), not pn. While this is only makes a slight
* different for large n, the formula used by RATS gives the "correct" empirical
* values for small n.
*
* The %FRACTILES function returns a 1-vector in this case, with the .05 quantile.
*
compute fr=%fractiles(intel,||.05||)
disp "5% VaR on 10000000, Empirical" -fr(1)*10000000