Dominance of a class of Stein type estimators for optimal portfolio weights when the covariance matrix is unknown (Q2268394)

The authors extend the class of Stein type estimators for the mean-variance optimal portfolio selection proposed in \textit{H. Mori} [J. Jap. Stat. Soc. 34, No.~1, Jap. Iss., 27--46 (2004; Zbl 1061.62173)] as follows. First, they introduce a general class of estimators given by Baranchik. Second, the authors consider estimators that shrink not only toward the origin but also toward an arbitrary fixed point. They show dominance results for the broader class of Stein type estimators when the covariance matrix is unknown and is estimated.