The concept of a synthesizing effect portfolio optimization (SEPO) model is proposed in this paper to study the investment portfolio problem for the first time. The SEPO model is a crisp programming model and obtained from a class of stochastic programming problems by constructing a class of synthesis effect functions. The SEPO model can further be shown to contain expectation value model by choosing different synthesis effect functions. A synthetically improved genetic algorithm based on real coding and random simulation is used in an illustrative example. It shows that the solutions of the SEPO model are richer than other solution models, and can be aware of different decision making in real life.