Efficient Algorithms for Conducting Stochastic Dominance Tests on Large Numbers of Portfolios: Reply

Abstract

In their comment on a paper by Porter, Wart, and Ferguson [7], Professors Frankfurter and Phillips [2] have raised two serious objections to “any attempt to compare and contrast SD and EV efficiency criteria on empirical grounds.” Essentially, they have argued that empirical comparisons of SD and EV selection rules are invalid because:a. The EV portfolio building algorithms are not allowed to operate in such tests, andb. While the estimators of true E and V are derived from and supported by elementary sampling theory, there exists no comparable sampling theory for the estimation of total probability functions.