Indeed the risk/return for the entire portfolio would have to be recalculated after adding back the newly weighted assets.
The minimum and maximum constraints work by ignoring portfolio iterations that fall outside of the prescribed constraints. We have just tested this with one product at both Min and Max of 10% which held for the optimal portfolio. If there are too many constraints (i.e. on other assets), it may be impossible to find weightings with a better return/risk profile. Could this be the case?
