Judge the combination, not just each holding
An investment’s standalone return estimate does not explain its entire portfolio role. Its variation and relationship with other holdings matter too. An exposure with a lower expected return can still change the tradeoff available to a combination.
This is the central shift from asking which single investment looks best to asking how different investments interact. It does not mean every additional holding improves a portfolio. A redundant exposure or an unfavorable price can still be unhelpful.
A simple modeled comparison
Consider three fictional feasible portfolios. A has an expected return of 4% and a modeled annual volatility of 6%. B has 5% and 6%. C has 6% and 10%. Under those estimates, A is dominated by B: B offers a higher expected return at the same modeled volatility.
B and C present a tradeoff. Selecting between them requires preferences and constraints that the table alone does not supply. The efficient frontier describes feasible combinations that are not dominated in this modeled sense; it does not promise a realized return or identify one universally correct portfolio.
The frontier is only as useful as its inputs
Expected returns, variances and covariances must be estimated. Changing them can move the frontier and change the suggested weights. Constraints on borrowing, short positions, position sizes and available assets also change which portfolios are feasible.
A mathematically precise optimizer can therefore produce economically fragile results. If a tiny change in an uncertain input causes an extreme change in weights, that sensitivity deserves attention. Precision of calculation should not be confused with certainty about the future.
Risk is broader than one model
Mean–variance analysis summarizes risk using variation and relationships among returns. Investors may also care about permanent losses, liquidity needs, taxes, extreme outcomes and changing relationships. A model should be evaluated against the question it is designed to answer.
The practical lesson is to inspect portfolio-level behavior and stress-test assumptions. The theory offers a language for comparing combinations. It does not make an optimized backtest a personal allocation recommendation or remove the need to understand the underlying exposures.
If Portfolio A and B have identical modeled volatility but B has higher expected return, which is dominated under those estimates?
Show the explanation
A is dominated by B within that model. This conclusion depends on the estimates and constraints, and does not guarantee B’s realized outcome.
Sources and editorial notes
Original explanations and hypothetical examples prepared with AI assistance. Primary references checked October 16, 2026. Numerical examples are hypothetical unless explicitly identified otherwise.
- CFA Institute: Portfolio Risk and Return Part I
Portfolio selection, efficient frontier and the role of risk and return estimates.