Standard deviation measures how widely a set of returns is dispersed around its average. In investing it is the most common proxy for volatility, and it treats upside and downside movement identically.
Dispersion. A higher figure means returns have been more spread out from their average; a lower one means they have clustered more tightly.
An unusually good month increases standard deviation exactly as much as an equally unusual bad one. No investor experiences those two events as equivalent, which is the first and most obvious limitation.
The measure is most informative when data is normally distributed. Market returns have fatter tails than a normal distribution — extreme events happen more often than the model implies — so standard deviation systematically understates how frequently large moves occur.
Calculated from a chosen historical window, and the answer depends on which window. A figure computed over a calm period describes the calm period.
It is comparable, cheap to compute, and standard across the industry, which makes it useful for comparing two things measured the same way. It is a poor summary of what a specific investor is exposed to, and drawdown usually communicates better.
Read next
What a risk score does not measure →