The idea is to add some kind of "statistical significance" check.
The "statistical
significance" (the higher the better) is proportional with the number of
trades (the higher the better) divided by the number of optimized parameters (the
lower the better). There should be some validation parameter which takes into account the number of optimized parameters in a given strategy and the number of resulting trades in IS/OOS period. This check could prevent from curve fitting / over optimization of the resulting strategies which are based on too many input parameters (= too many degrees-of-freedom). So taking into account the number of degrees-of-freedom used for the strategy generation, this mechanism could give users a warning that the number of trades is probably too low to be considered statistically significant.