What is the global minimum variance portfolio in a multi-asset setting, and what are its unique properties?

Question

AcadiFi · Accepted Answer

The global minimum variance (GMV) portfolio generalizes the two-asset MVP to any number of assets and has properties that make it uniquely attractive to risk-averse investors. **Definition** The GMV portfolio is the portfolio with the **lowest possible variance** among all portfolios formed from the available asset universe. It sits at the **leftmost point** of the minimum-variance frontier (the full parabola, not just the efficient upper half). **Multi-Asset Formulation** For n assets with covariance matrix Sigma and a vector of ones (1), the GMV weights are: **w_GMV = (Sigma^-1 * 1) / (1^T * Sigma^-1 * 1)** This is a matrix operation — the inverse of the covariance matrix, multiplied by a vector of ones, then normalized to sum to 1. **Worked Example — Thornbury Three-Asset GMV:** Three assets with the following covariance matrix (annualized): | | Ridgeway Equity | Holbrook Bonds | Windmere REIT | |---|---|---|---| | Ridgeway Equity | 0.0625 | 0.0075 | 0.0120 | | Holbrook Bonds | 0.0075 | 0.0100 | 0.0030 | | Windmere REIT | 0.0120 | 0.0030 | 0.0400 | After computing Sigma^-1 and applying the formula: | Asset | GMV Weight | |---|---| | Ridgeway Equity | 8.2% | | Holbrook Bonds | 77.4% | | Windmere REIT | 14.4% | | **GMV Sigma** | **9.18%** | **Unique Properties of the GMV Portfolio:** 1. **No expected return inputs needed.** The GMV depends only on the **covariance matrix** — no return forecasts required. This makes it robust to estimation error in expected returns (the most unreliable input in mean-variance optimization). 2. **Maximum Sharpe ratio portfolio it is NOT.** The GMV minimizes risk without regard to return. The tangency portfolio (maximum Sharpe ratio) is a different point on the frontier. 3. **Beta of every asset with respect to the GMV portfolio equals 1.** This is a mathematical identity: in the GMV portfolio, all assets contribute equally to marginal risk. 4. **Out-of-sample performance.** Empirically, the GMV portfolio often **outperforms** the tangency portfolio out of sample because it avoids the large estimation errors in expected returns that plague mean-variance optimization. \`\`\`mermaid flowchart TD A["Asset Universe
(n assets)"] --> B["Estimate Covariance
Matrix Σ"] B --> C["Compute Σ⁻¹"] C --> D["w = Σ⁻¹ · 1 / (1ᵀ · Σ⁻¹ · 1)"] D --> E["Global Minimum
Variance Portfolio"] E --> F["Lowest σ on
Entire Frontier"] E --> G["No Return
Estimates Needed"] E --> H["All Betas = 1
w.r.t. GMV"] \`\`\` **Why Practitioners Use It:** - **Estimation error robustness:** Expected returns are notoriously difficult to forecast. The GMV avoids this input entirely. - **Conservative mandate:** For risk-averse clients who prioritize capital preservation over return maximization. - **Benchmark for risk budgeting:** The GMV serves as a reference point — any deviation from GMV is a deliberate risk bet. **Exam Context:** Know the formula, understand that it requires only the covariance matrix, and be able to explain why it often outperforms more "optimal" portfolios in practice. Explore multi-asset optimization in our CFA Level II Portfolio Management course.

What is the global minimum variance portfolio in a multi-asset setting, and what are its unique properties?

Master Level II with our CFA Course

Related Questions

Practice Questions