Using a genetic method when optimizing a strategy is preferred if the number of iterations becomes too many to be finished within any reasonable time.
The genetic optimizer starts by setting up:
Each population is a randomly selected combination of parameters.
Then the optimizer starts changing the parameters of the populations for each generation. The parameters are shifted randomly, and for each generation, the worst parameter combinations are thrown away, and the better ones are kept.
Then the optimizer keeps on repeating this step for the set number of generations, and returns the best parameters that it encountered during the process.
The process is thought to resemble that of natural selection.
The fact that populations are thrown away during the optimization, means that the number of iterations decreases as the optimizer gets closer to finishing. That’s why a genetic optimization usually accelerates during the optimization.
If the number of iterations becomes too many and will take too long to optimize, then you could go with genetic algorithms.
A good rule of thumb is to try to cover at least 10% of the optimization space.
That means, that if you have 10000 iterations for a strategy when using the exhaustive option, you should make sure that the generations times the populations is at least 1000.