Abstract: Two portfolio selection models,the traditional Markowitz Mean-variance model and Minimax model,proposed by Young in 1998,are studied and empirically compared in this paper.The true historical data of Chinese Shangzhen 30 Index is divided into two parts:one as sample data for portfolio optimization analysis,the other as out-of-sample data for testing.The results show that risk-return efficient frontiers have very similar shapes.But the performance of strategy based on Minimax model are better than that of MV model.The empirical results prove the theoretical conclusions of the paper and show that Minimax is very operational and practicable.

Key words: portfolio selection, mean-variance model, minimax model, measure of risk