Comparing investments by return alone rewards recklessness, because risky things produce impressive numbers on the days they work. The correction is obvious once stated: ask how much extra return arrived per unit of turbulence endured. That ratio – excess return above what safe government debt pays, divided by how bumpy the ride was – carries William Sharpe’s name and remains the most-used summary statistic in professional money management, which is precisely why its limitations deserve more attention than its formula.
Written out: take average return, subtract the return available from near-certain government borrowing over the same period, divide by standard deviation of returns. Higher means better compensation for discomfort. Every term contains a judgment call, starting with what counts as the risk-free benchmark – short Treasury bills conventionally, though nobody checks whether the chosen window flattered whichever figure appears beside it. Standard deviation deserves suspicion too, since it treats every departure from average symmetrically – volatility gets examined honestly elsewhere: an unexpected gain and an unexpected collapse each widen the denominator equally, penalising upside you would happily have kept.
Two funds both returning ten percent annually make the point cleanly. One drifted upward in a smooth line; the other doubled down to eighty and back twice, experiencing stretches where anyone holding it questioned their judgment permanently. Same headline, incomparable experiences, and only the ratio exposes the difference – because most investors abandon sound plans during rough patches rather than during boring ones, realised behaviour tracks the second column and not the first. Choosing higher readings within your own strategy family genuinely helps. Comparing across strategies does not, for four structural reasons worth knowing before quoting any score.
Distribution shape is invisible. Standard deviation presumes deviations cluster predictably around a middle value. Actual markets produce fat tails: rare events far beyond what normal-looking history suggests, arriving in clusters just when holdings concentrate together. A strategy collecting modest gains for years and losing catastrophically once can post a comfortable reading right up until the moment it doesn’t, and nothing in the calculation warns you, since the event defining the strategy hasn’t appeared in the measurement window yet. Insurance-style return streams describe this pattern precisely, which is how some funds historically looked brilliant before ceasing to exist.
Timeframe selection manufactures results. Measure a bond fund alongside a growth fund over rising-rate years and the comparison becomes nonsense; extend windows and rankings shuffle routinely. Anyone displaying a favourable figure chose the window, cadence of observations and length, and there are enough combinations that almost any product’s ranking can be made attractive. Ask immediately what period produced this, whether it includes a genuine downturn, and whether excluding the last few years changes the winner – it frequently does.
Correlation and tail co-movement aren’t captured. A portfolio combining several assets scoring well individually behaves differently when they move together under stress than when they don’t, and individual ratios carry no information about joint behaviour. Portfolio quality depends substantially on relationships between holdings that no per-holding statistic can see, which makes aggregation of good scores a false shortcut – allocation depends exactly on this invisible layer.
Costs, taxes and access vanish. Gross figures look attractive before fees, spreads and account restrictions consume returns; illiquid holdings produce smoothed valuations that suppress apparent variability and therefore inflate scores mechanically, an artifact professionals know to distrust whenever private markets appear on comparison tables.
Used well the metric answers a narrow question: within a category of similar approaches, which delivered comparable performance with less thrashing? That comparison is genuinely useful – screening five index-tracking vehicles or comparing two managers pursuing identical strategies. Used badly it ranks unrelated products into a league table, encouraging substitution toward whatever measured smoothly recently, which is one documented pathway by which ordinary investors end up holding yesterday’s calmest asset immediately before it becomes today’s worst.
Practical substitutes often beat the ratio outright: maximum drawdown asks how bad it got before recovery, which maps to whether you’d have kept owning; sequence-of-returns matters more than averages for money being withdrawn, a subject covered in the retirement mathematics; required return comparisons against alternatives you’d hold instead keep the analysis grounded, since accepting extra wobble only makes sense when the extra payment is real relative to the boring option – that trade-off deserves its own framing. And remember the underlying purpose of measuring at all: not building leaderboard habits but calibrating how much turbulence you can sit through without touching the sell button, which is where most of the actual damage in ordinary portfolios occurs, consistently, at exactly the wrong hour. Nobody improved their results by finding a higher score; plenty improved them by holding something less exciting they could actually keep.
This is educational explanation of a performance statistic, not financial advice or a recommendation of any product; no funds are named because rankings shift with the measurement window and quoted figures go stale. Investing involves risk: illustrative numbers are invented round values supplied for checking arithmetic, past measurements do not predict future behaviour, every measure rests on assumptions a given security may violate, and any holding can lose value however smoothly it once performed.