Expected Utility Regret Rule: Minimax and Bayes Optimal Portfolio Choice

Battle Count: 50

Highly relevant for institutional asset allocation and systematic portfolio construction. It provides a rigorous theoretical foundation for when simple heuristics like Mean-Variance or Risk Parity are optimal and quantifies the cost of estimation error in portfolio selection.

The theoretical framework is complex, involving minimax/Bayes analysis and linear programming for class selection. However, the implementation for specific cases (like Mean-Variance or Risk Parity) is straightforward. The general EUR rule requires solving a finite linear program for class selection when more than two classes are candidates.

The paper provides detailed algorithms (Algorithm 1), specific parameter settings for simulations, and uses standard public data (French Data Library) for empirical studies. Proofs are included in appendices.