Mean Variance Optimization (MVO), pioneered by Harry Markowitz, is the mathematical engine behind the Efficient Frontier. Given a set of assets, their expected returns (means), variances, and pairwise correlations (covariances), MVO finds the set of portfolio weights that produces the best risk-return tradeoff.
The key inputs are: (1) expected return for each asset, typically estimated from historical returns; (2) the covariance matrix capturing how assets move together. MVO then solves a quadratic programming problem to identify the optimal weight vector for a target return or risk level.
MVO's main weakness is its extreme sensitivity to input assumptions. Small errors in expected return estimates can produce wildly different weight recommendations. A portfolio that is 'optimal' based on historical data is rarely optimal going forward — this is known as 'error maximization.' Robust alternatives like Hierarchical Risk Parity address this sensitivity.
On StressTest.pro, MVO is available under the Optimizations tab. Results should be treated as a starting point for portfolio construction, not a definitive prescription.