Parameter Stationarity and the Mean Variance Optimum Portfolio
Seidman College of Business
Most procedures for selecting weights for mean variance optimal portfolios are based upon historical variances and covariances. One well known but little discussed issue associated with predicting future (ex anti) mean variance weights by using historical (ex post) parameters is the extent to which the historical parameters predict future parameters. This is the issue of parameter stationarity. In the following discussion EW will refer to the set of equal weights (i.e. 1/n) and MV will refer to the set of weights derived by using historical parameters and mean variance optimization.
Blose, Laurence and Griggs, Frank, "Parameter Stationarity and the Mean Variance Optimum Portfolio" (2013). Faculty Scholarly Dissemination Grants. 1256.