Universal Portfolios Generated by Vandermonde Generating Matrix

A universal portfolio generated by the one-parameter symmetric positive definite Vandermonde matrix is studied. It is obtained by maximizing the scaled growth rate of the estimated daily wealth return and minimizing the Mahalanobis squared divergence of two portfolio vectors associated with the Vandermonde matrix. The parameter of the Vandermonde matrix is chosen so that the matrix is positive definite. The companion matrices of the three and five-dimensional generating matrices are evaluated to determine the portfolios. Three and five stock-data sets are selected from the local stock exchange in Malaysia and the empirical performance of the portfolios is presented. There is empirical evidence that the use of an appropriate generating Vandermonde matrix may increase the wealth of investors.