The portfolio that returned more isn't the better one: risk-adjusted return
4 min read by Opthest
Put two of them side by side. Over the same decade, portfolio A returned 8% a year, B returned 6%. Easy question: which is better? The instinctive answer — A, obviously — is the same one that makes one investor in two choose badly. Because half the information is missing: how much they shook to get there. If A dragged you through 40% crashes and sleepless nights while B fell an orderly 12%, “returned more” stops being an answer. Return alone is a half-truth. The other half is risk — and there’s a way to put them on the same scale.
What does the Sharpe ratio actually measure?
Return per unit of risk, nothing more esoteric. You take the portfolio’s return, subtract what a risk-free asset paid (a short-term government bond: what you’d have earned without risking anything), and divide the result by volatility — the average size of the swings. The numerator is the premium the market paid you for taking on risk; the denominator is how much risk you took to collect it. A Sharpe ratio of 1 means, roughly, one point of excess return for every point of turbulence. The higher it is, the more efficient the trade was.
Why subtract the “risk-free” return?
Because part of the gain you didn’t earn — time handed it to you. If government bonds paid 3%, a portfolio that made 4% added a meager single point of premium for all its agitation — not four. Subtracting the risk-free rate isolates the only thing worth judging: the reward for the risk you chose to run. It’s the difference between asking “how much did I make?” and “how much was I paid for the trouble?”. Only the second question says anything about the quality of the choice.
So is a high number always better?
Almost — but with two traps it’s only honest to name. First: Sharpe treats volatility as a synonym for risk, and counts swings up and down the same way. A portfolio that occasionally spikes upward gets punished for the very restlessness that made it money — which doesn’t always make sense. Second, and more serious: volatility measured on ordinary data understates disasters. Markets have fat tails — rare but violent crashes that a “smooth” measure doesn’t see coming — and for exactly that reason a Sharpe ratio can look reassuring right up to the day before the wreck. It’s an excellent ruler as long as you’re measuring regular cloth; less reliable on sudden jumps.
Over what period am I reading it?
A decisive question, and almost always forgotten. The Sharpe ratio is not a constant carved into the portfolio: it’s a snapshot of a specific period. Computed over a bull market it looks splendid; recomputed to include the year of the crash it deflates. The conventions shift too — you normally “annualize” from monthly or daily data, and a different method moves the number. That’s why a Sharpe ratio only means something if you know over which window it was measured and you compare it with another computed the same way. It’s the same caution as a backtest: a number without its time interval is an anecdote dressed up as a metric.
So how do I actually use it?
As a comparison, never as a target. The Sharpe ratio doesn’t tell you which portfolio to buy: it tells you which of the two paid better for its risk in that past, leveling the field so A and B are genuinely comparable. It’s no accident that it sits at the heart of the efficient frontier: the “tangency” portfolio, the one that maximizes return per unit of risk, is exactly the highest Sharpe ratio achievable given the assumptions. But it stays a compass, not a map: it points to an efficient direction, guarantees nothing about the future, and doesn’t replace the question that matters — how much risk you’re willing to bear. That one stays yours.
Return tells you how far you got. Sharpe tells you how rough the road was to get there. Judging a portfolio by the first alone is like picking a flight by its arrival time and ignoring the turbulence.