Mean-Variance Optimization, the Efficient Frontier, and Estimation Risk

Summary

This tutorial develops the Markowitz efficient frontier from mean-variance analysis, describing the minimum-variance portfolio, the maximum-Sharpe portfolio, and combinations that form the efficient set. It explains how forecasts of asset returns and a covariance matrix can guide portfolio weights, rather than relying only on equal weighting or ranking assets by predicted return. Simulated examples compare an equal-weight selection strategy with a Sharpe-oriented portfolio and describe a historical-data backtest using a small Chinese equity universe.

The tutorial emphasizes that frontier estimates are sensitive to uncertain expected returns and covariances, with estimation error particularly affecting the tangency portfolio and capital market line. It presents a Monte Carlo illustration and reports that the optimized portfolio had higher Sharpe ratios in one simulation and higher returns in the described backtest, but these are examples rather than general evidence. The framework assumes risk aversion and mean-variance preferences; practical use also faces nonstationary data, transaction costs, short-sale constraints, leverage limits, and distributional assumptions that may not hold.

Key ideas

• The efficient frontier contains portfolios that are mean-variance optimal for a given level of risk.
• Portfolio weights depend on expected returns and the covariance matrix, not just asset rankings.
• The article illustrates how equal-weight selections can fall away from the frontier even when a return forecast is used.
• Estimation errors can substantially shift the frontier and the maximum-Sharpe portfolio.
• Transaction costs, short-sale and leverage constraints, and return-distribution assumptions limit practical application.

Tags

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.