Robust Mean-Variance
Worst-case mean-variance optimisation over an ellipsoidal uncertainty set on expected returns. Penalises portfolios that bet too aggressively on noisy point estimates of .
Overview
Plain mean-variance optimisation is famously fragile because it treats the sample mean of returns as if it were the truth. Robust MVO recognises that is itself a noisy estimator and instead solves for the portfolio that performs best against the worst case mean inside an ellipsoidal uncertainty set around .
The result penalises positions whose Sharpe is sensitive to mean estimation error, naturally producing more diversified, less concentrated allocations than vanilla MVO.
Mathematical Formulation
Let be the sample mean, the (Ledoit–Wolf shrunk) covariance, and the covariance of the mean estimator. For daily data with observations and annualised covariance :
The robust counterpart is:
where scales the ellipsoidal uncertainty radius. Folio Lab uses , , and risk-aversion by default. The problem is a second-order cone program solved with CLARABEL or SCS.
Why This Helps
Vanilla MVO is highly sensitive to errors in — small perturbations in the mean produce large weight swings and corner solutions. The added term is precisely the worst-case loss of the linear functional over an ellipsoid, so the robust objective is exactly the lower bound of the expected return that holds for every in the uncertainty set.
Advantages & Limitations
Advantages
- Estimation-error aware: Hedges against noisy .
- Convex SOCP: Tractable with standard solvers.
- Less concentrated: Naturally produces diversified portfolios.
- Stable weights: Lower turnover than vanilla MVO.
Limitations
- Conservative: Can give up upside when is actually accurate.
- Hyperparameter choice: Confidence and uncertainty aversion are user picks.
- Still trusts the covariance: No protection against errors in .
References
- Ceria, S., & Stubbs, R. A. (2006). "Incorporating estimation errors into portfolio selection: Robust portfolio construction." Journal of Asset Management, 7(2), 109-127.
- Goldfarb, D., & Iyengar, G. (2003). "Robust portfolio selection problems." Mathematics of Operations Research, 28(1), 1-38.
- Fabozzi, F. J., Kolm, P. N., Pachamanova, D. A., & Focardi, S. M. (2007). Robust Portfolio Optimization and Management. Wiley.