The minimum variance portfolio: construction, trade-offs, and when it outperforms

The minimum variance portfolio is the portfolio on the efficient frontier with the lowest possible volatility. Unlike the tangency portfolio—which maximises the Sharpe ratio and requires expected return estimates—the minimum variance portfolio requires only an estimate of the covariance matrix of returns. This makes it robust to one of the most practically difficult inputs in portfolio optimisation and has led to its widespread use as an alternative to market-cap weighting in factor investing and smart beta strategies.

What the minimum variance portfolio is

In Markowitz's (1952) mean-variance framework, the efficient frontier is the set of portfolios that offer the highest expected return for each level of volatility. The minimum variance portfolio sits at the leftmost point of this frontier—it has the lowest volatility of any combination of the available assets and cannot be made less volatile by changing the weights. Every other portfolio on the efficient frontier has higher expected volatility; whether it also has higher expected return depends on the specific assets and the investor's return forecasts.

The mathematical construction of the minimum variance portfolio requires solving a quadratic optimisation: minimise portfolio variance (w'Σw, where w is the weight vector and Σ is the covariance matrix) subject to the constraint that weights sum to one and, in a long-only implementation, that all weights are non-negative. The solution is determined entirely by the covariance matrix—no expected return inputs are required. This is the key practical advantage: expected return estimates are notoriously difficult to forecast accurately and introduce substantial estimation error into the tangency portfolio optimisation. The minimum variance portfolio sidesteps this entirely.

What the evidence shows

Clarke, De Silva, and Thorley (2006, 2011) studied the performance of minimum variance equity portfolios constructed from the US stock universe over 1968–2009 and found that these portfolios delivered returns comparable to the broad market while exhibiting significantly lower volatility—typically 25–30% lower standard deviation. On a risk-adjusted basis, minimum variance portfolios achieved substantially higher Sharpe ratios than cap-weighted benchmarks. This finding has been replicated across global equity markets and is one of the most robust empirical results in quantitative equity portfolio construction.

The outperformance of minimum variance portfolios on a risk-adjusted basis is related to, but distinct from, the low volatility anomaly—the observation that lower-risk assets have historically delivered higher risk-adjusted returns than higher-risk assets. The minimum variance portfolio exploits this anomaly structurally by constructing a portfolio that specifically seeks low-volatility assets and low-correlation combinations, but the result is a portfolio property rather than a single-factor bet.

Limitations and trade-offs

The minimum variance portfolio is sensitive to the covariance matrix used as input. If the estimated covariance matrix differs significantly from the true future covariance structure—which it often does, since covariances are themselves volatile and regime-dependent—the resulting portfolio may not actually achieve minimum variance in practice. Long estimation windows (three to five years) produce more stable covariance estimates but may lag structural changes in correlation regimes; short windows are more current but noisier.

Minimum variance portfolios frequently concentrate in a small number of assets with the lowest volatility and highest negative correlations—typically low-volatility sectors such as utilities, consumer staples, and healthcare. Without explicit diversification constraints, the resulting portfolio can be uncomfortably concentrated compared to a broad market index. Position limits and sector caps are commonly applied to minimum variance portfolios in practice to prevent this concentration, but they move the solution away from the true mathematical minimum variance.

The minimum variance portfolio does not maximise returns—it minimises risk. In sustained bull markets, a minimum variance portfolio will typically underperform a high-beta or equally-weighted market portfolio because it underweights the assets with the strongest upside. The trade-off is a smoother return path with smaller drawdowns, which the evidence suggests produces a better Sharpe ratio over full market cycles but requires patience through extended periods of relative underperformance.

Minimum variance portfolio in pfolio

pfolio offers three portfolio optimisation methods: mean-variance optimisation (MVO), Hierarchical Risk Parity (HRP), and equal weight. Mean-variance optimisation on a risk-minimisation objective is related to minimum variance construction—both use the covariance matrix as a central input—while HRP provides an alternative approach to covariance-based optimisation that is more robust to estimation error. Investors can select and compare optimisation methods in the Portfolios section at pfolio.io/portfolios. Methodology details are available in how we build portfolios.

Related articles

• Mean-variance optimisation: how to find the highest Sharpe ratio portfolio
• The tangency portfolio: the highest Sharpe ratio on the efficient frontier
• Hierarchical risk parity: a more robust approach to covariance-based portfolio construction
• The low volatility anomaly: why lower-risk assets have often outperformed higher-risk ones