322 THE HANDBOOK OF PORTFOLIO MATHEMATICS
infinity—that is, as you use a larger and larger value for unimproved
generations—the answer converged upon is exact. However, both of these
parameter increases are at the expense of extra computing time.
The algorithm can be time intensive. As the number of scenario sets
increases, and the number of scenarios increases, the processing time grows
geometrically. Depending upon your time constraints, you may wish to keep
your scenario sets and the quantity of scenarios to a manageable number.
The genetic algorithm is particularly appropriate as we shall see by Chapter
12, where we find the landscape of leverage space to be discontinuous for
our purposes.
Once you have found the optimal portfolio, that is, once you havef
values, you simply divide thosefvalues by the largest loss scenario of
the respective scenario spectrums to determine thef$ for that particular
scenario spectrum. This is exactly as we did in the previous chapter for
determining how many contracts to trade in an optimal portfolio.