Guided lab · synthetic uncertainty set
What does κ actually buy you? Move the ellipsoid radius and watch the worst-case mean penalty change the allocation. The optimizer cannot rescue invalid covariance or contaminated price history, so the failure scenario stops before optimization.
Scenario Canonical valid inputs Comparison: κ = 0 Failure: covariance is not PSD
Uncertainty radius κ: 0.50
Back Step Play Pause Reset
Mean uncertainty ellipse and robust objective An uncertainty ellipse grows with kappa. An arrow points from nominal mean to the adverse boundary, and a two-asset allocation bar reports the robust optimum. μ̂ adverse mean for w Robust allocation worst return 5.804% variance 1.093% objective 4.710% A 44.21% · B 55.79% Stage 1 of 4 · Validate the nominal mean and covariance. Live decomposition Weight A44.21%
Weight B55.79%
Mean penalty0.523%
Objective4.710%
μ̂ᵀw − κ‖Qᵀw‖₂ − λwᵀΣw
Canonical κ = 0.50 loaded. Press Step to reveal the contract.
Keyboard: Right/Space steps, Left goes back, R resets, P plays or pauses. Reduced-motion mode disables autoplay.