Traditionally parsimony pressure has been used with genetic programming to reduce the complexity of solutions analogous to the principle of Occam's Razor. But there have been several signs from previous experiments that this reduces the quality of the solutions. In an attempt to counteract this we present one of the first experiments that try to apply negative parsimony pressure on genetic programming, ie. we prefer complex solutions rather than simpler ones. This system is then applied on a financial portfolio optimization problem to test it's performance on real world data. Our results indicate that negative parsimony pressure work better than regular parsimony pressure on average, and it's almost always better to use some kind of parsimony pressure than not.