The purpose of this article is creating the suitable model for investment. First, according to the literature review, we extract the effective criteria for portfolio selection. Then we use experts' opinions about the importance of each criterion to find criteria pair wise comparisons. In this stage, because of interrelationship among the criteria, ANP is used as the most appropriate technique for this analysis. At the next step, we use TOPSIS technique for ranking the population members, including companies of seven industries (cement, main mineral, petrochemicals, medicines, real estate, automobile). Real data average of three years – from 2007 to 2009 – is used as input for TOPSIS technique. After determining the companies ranking, the selected portfolio includes 40 top companies. In order to confirm the selected portfolio, the three years return of the selected companies is compared with non selected companies. After benchmarking portfolio based on Sharp and Treynor criteria, we use Memetic algorithm to optimize the selected portfolio.