Comparative Studies of Stochastic Portfolio Construction Models with Geometric Average Rate of Return

摘要

Every investment has the return and the risk. The portfolio needs to be determined to maximize the return and to minimize the risk. This paper compares two stochastic linear programming models to construct investment portfolio; mean-lower semiabsolute deviation (LSAD) model and Conditional Value-at-Risk (CVaR) constraint model. As a performance measure for investment decision when scenarios for only a single period are known, geometric average rate of return (GARR) is proposed. The comparative studies show that GARR is a good risk-adjusted rate of return for single-period investment problems.