A novel approach on the portfolio selection theory is given with regard to advanced utility performance that incorporates more accurate investor patterns up to the fifth moment. Bankruptcy detection, a priori, on an investment portfolio of stocks is a significant process that can eliminate potential losses. Even in case of corporate fraud, efficient funds can maximize their net present value by reforming the assets. Multi-layer perceptron neural networks are compared with hybrids of neuro-genetic multi-layer perceptrons and the voted-perceptron algorithm to define the most efficient classification method into the perceptrons family, implementing extensive network topologies.