The efficient frontier is the set of portfolios that offer the highest expected return for each level of risk. Any portfolio below the frontier could be improved by earning more for the same risk or taking less risk for the same return.
Reading the frontier
Plot risk (volatility) on the horizontal axis and expected return on the vertical axis. Possible portfolios fill an area, and its upper-left edge, from the minimum-variance portfolio upward, is the efficient frontier.
An illustrative example
Using the two assets in mean-variance optimization (A: 10% return, 20% volatility; B: 6% return, 10% volatility; correlation 0.3):
| Mix (A / B) | Expected return | Volatility |
|---|---|---|
| 0 / 100 | 6.0% | 10.0% |
| 20 / 80 | 6.8% | 9.96% |
| 50 / 50 | 8.0% | 12.4% |
| 100 / 0 | 10.0% | 20.0% |
Holding B alone is not efficient: the 20/80 mix has slightly lower risk and a higher return. The least-risk mix is about 10.5% A and 89.5% B, at about 9.8% volatility. The numbers are made up.
The tangency portfolio
With a risk-free asset, the best frontier portfolio is the one with the highest Sharpe ratio, the tangency portfolio. Investors then mix it with cash to set their level of risk.
Limits
- The frontier is drawn from estimated inputs, so it is uncertain and unstable. An estimated frontier is always better than the true one in-sample.
- Variance treats all deviations as risk and ignores fat tails. See expected shortfall.
- Real portfolios face constraints and costs. See optimization constraints and portfolio optimization.
Frequently asked questions
What is the efficient frontier?
It is the set of portfolios that give the highest expected return for each level of risk, or the lowest risk for each level of return.
What is the tangency portfolio?
The portfolio on the efficient frontier with the highest Sharpe ratio, which investors combine with a risk-free asset.
Why are portfolios below the frontier inefficient?
Because another portfolio offers more return for the same risk, or less risk for the same return.
Is the efficient frontier reliable in practice?
It depends on estimated returns and covariances, which are uncertain, so the computed frontier can be unstable and over-optimistic.