Abstract: There are two methodological flaws in conventional approach on ranking mutual fund. First, the real way of the investor making decision is neglected and it is assumed that the investor has the same aptitude on gain and loss. Second, the properties of empirical sample are ignored and it is assumed that the sample of stochastic returns reaches the size of asymptotic normality population. Based on expectation utility theory, the excess returns are expanded by high order Taylor series to link the high order moments of excess returns and utility function. Using higher order moments as constraints, the empirical probability is estimated from sample and adjusted by decision weight. Based on empirical probability decision adjustment, mutual funds are ranked by Sharpe ratio expansion and stochastic dominance criterion. As a case study, 645 mutual funds are rated with this two indicators.

Key words: probability weighting function, stochastic dominance, mutual funds