We present in this paper a new approach to portfolio selection using imprecise model, Fuzzy goal programming and Genetic algorithm. The results of this approach is compared with traditional portfolio selection and presented here. In conventional methods, the skewness of portfolio returns is not considered. In this new approach, we prefer the stocks in the portfolio which has demonstrated positive skewness in the past patterns. By careful analysis of past patterns we give preference to positively skewed stocks and strengthen the portfolio. The paper discusses a brief literature survey and elaborates on our model and finally gives the results and conclusions.