AbstractGiven a liability structure, the bank portfolio optimization determines an asset allocation that maximizes profit, subject to restrictions on Basel III ratios and credit, liquidity, and market risks. Bank allocation models have not been tested using historical data. Using an optimization model based on turnover constraints, we develop such tests, which document the superior performance of optimization strategies compared to heuristic rules, resulting in an average annual out‐of‐sample outperformance of 15.1% in terms of return on equity using our data set. This outperformance is remarkable and contrasts with the reported underperformance of several portfolio optimization methods in the case of investment management.