The mean variance optimization (MVO) approach has been used widely in investment practice, however, it contains many shortcomings. First it assumes that investors solve one period investment problem, when in reality many investors are concerned about inter-period volatility. Additionally, MVO assumes that expected return and expected risk inputs are known with certainty. In fact, these input parameters are unknown, and any use of these inputs will involve estimation error. To address the single period issue inherent in MVO, we assume that investors have both long- and short-term return and risk expectations. To address the estimate error issue, we apply insights from the Tütüncü and Koenig (2004) work on robust optimization in our stylized dual investment horizon setting. This allows for a rich analysis of portfolio choice, solving a more realistic investor problem.