None of the models that have been developed to determine the optimal strategic asset allocation (SAA) of stabilization sovereign wealth funds (SWFs) has received direct empirical validation, primarily because there is a lack of transparency regarding some of the key parameters that characterize the problem. In this paper, building on a mean‐variance framework, we derive three sets of parsimonious statistical tests to compare the actual SAA of SWFs to a theoretical optimum. We apply these tests to the portfolio of the world's largest stabilization SWF (the Norwegian Government Pension Fund—Global or GPF) for the period between 2002 and 2005. The empirical analysis confirms that the static and dynamic deviations of the GPF's SAA from the market equity portfolio are consistent with the theoretical predictions.