SUMMARY
This study aims to determine the difference of return and risk candidate and non-candidate optimal portfolio. The population of this research is company member LQ 45 period February 2009 to January 2014 sampling technique was purposive sampling, and acquired 20 companies in the sample. Optimal portfolios are measured using a single index model. The hypothesis was tested using two different test average using the Mann-Whitney analysis model (U-test). The results showed 20 LQ-45 continuously or at least 10 members appeared in LQ-45 stocks gained 5 candidate optimal portfolio. There were no differences between return and risk and non-candidate candidate optimal portfolio.