Ask most people whether one portfolio is better than another and they will compare returns. Modern Portfolio Theory, the idea Harry Markowitz formalized in 1952, says that question is only half-formed. A return figure means nothing until you know the risk taken to earn it — and how the holdings move together.
Every portfolio is a point
Fix a set of assets and you can build countless portfolios simply by changing the weights — 60/40, 70/30, an equal split across ten names. Give each portfolio two coordinates:
- Horizontal: its volatility — how much the value swings.
- Vertical: its expected return.
Now every allocation is a single dot. Generate thousands of weight combinations and the dots form a cloud. The chart below is exactly that: a deterministic sample of random portfolios, each one a dot placed by its own risk and return.
The edge that matters
Look at the cloud's upper-left boundary. For any level of volatility, one portfolio sits higher than all the others at that risk — it earns the most return for that much risk. Trace those best-in-class portfolios and you get the efficient frontier, the curved edge of the cloud.
Every dot inside the cloud is dominated: there is another portfolio with the same risk and more return, or the same return and less risk. Holding a dominated portfolio means leaving return on the table for no reason. The whole point of the exercise is to move onto the frontier.
Two portfolios worth naming
The frontier is a menu, not a single answer. Two points on it earn names:
| Portfolio | Volatility | Expected return | Return per unit of risk |
|---|---|---|---|
| Minimum-volatility | 34.7% | 27.6% | 0.79 |
| Maximum-Sharpe | 46.2% | 66.5% | 1.44 |
The minimum-volatility portfolio is the leftmost tip of the frontier — the calmest allocation you can build from these assets. It is not the highest return; it is the lowest risk.
The maximum-Sharpe portfolio is the point where each additional unit of risk buys the most extra return — the best return per unit of risk. Here it earns 1.44 units of return for every unit of volatility, against the minimum-volatility portfolio's 0.79. It takes more risk than the calm tip, but it is paid far better for doing so.
Weight is not risk
A subtle trap hides inside every allocation: how much capital you put in an asset is not how much risk it contributes. A 20% position in a volatile, highly correlated asset can easily carry half of the portfolio's total risk, while a 20% position in a calm, uncorrelated one carries almost none. Two portfolios with identical weights can have completely different risk profiles once correlation is accounted for. Judging a portfolio by its weights alone is judging it by the wrong number.
Correlation is the only free lunch
Volatility does not simply add up. When two assets fail to move in lockstep, their swings partly cancel, and the portfolio's volatility falls below the weighted average of its parts. The lower — or more negative — the correlation, the larger this cancellation.
Two measures make it concrete:
- The diversification ratio compares the weighted-average volatility of the holdings to the portfolio's actual volatility. The further above 1, the more the correlations are working in your favor.
- The effective number of bets asks how many genuinely independent positions you really hold. A ten-asset portfolio where everything moves together is, in risk terms, close to a single bet — not ten.
This is why diversification is often called the only free lunch in investing: it can lower risk without lowering expected return, purely by combining things that do not move together.
The reframe
Modern Portfolio Theory quietly changes the question. You stop asking "how much did it make?" and start asking "how much return did it earn per unit of risk — and which holdings are actually carrying that risk?" Every dot inside the cloud is a reminder that return without a risk denominator is an incomplete number.
That denominator is the subject of the metrics ahead in this series — the ratios that turn a return figure into a statement about the risk behind it.