Minimax: Portfolio Choice Based on Pessimistic Decision Making

We propose a fund allocation strategy for a highly risk-averse investor based on pessimistic decision making to construct portfolios of four major asset classes. Using US data (indexes of stocks, bonds, real estate, and commodities) from January 1990 to December 2010, we find that the proposed Minimax strategy performs well out-of-sample with respect to standard risk measures. Its performance is better than common alternative trading strategies such as fixed weights, minimum variance, or mean-variance methods. Portfolio weights are stable across time, resulting in lower turnover than any mean-variance related strategy. Finally, we find that optimal portfolios are widely diversified across all asset classes. This study suggests that the proposed Minimax strategy is implementable in portfolio management, especially for large institutional investors.