The Hidden Sharpe Ratio

We estimate the maximal Sharpe ratio in large cross-sections of assets directly, without constructing the efficient portfolio. Covariance estimation in this setting requires regularization, which suppresses the contribution of small-eigenvalue directions. Our estimator progressively recovers this contribution, yielding tight upper and lower bounds with asymptotic confidence intervals. We further show that ridge-regularized Markowitz portfolios, the construction underlying many existing high-Sharpe benchmarks, deliver out-of-sample Sharpe ratios that fall short of the true maximum by an explicit, non-vanishing multiplicative factor. Applied to U.S. equities, the method delivers an estimated maximal annualized Sharpe ratio of approximately 7.4, or 5.0 once micro-cap stocks are excluded, substantially exceeding existing portfolio-based benchmarks of 1.0–4.0. Adding corporate bonds raises the maximum, though not by a statistically significant margin. Nonlinear portfolio weights dominate linear ones, and characteristics introduced after 2004 contribute little additional Sharpe ratio.