Standard Deviation

The calculation proceeds in fixed steps. Compute the arithmetic mean of the observations, subtract it from each observation to obtain deviations, square each deviation so that positive and negative departures do not cancel, sum the squares, divide by the number of observations for a population or by observations minus one for a sample, then take the positive square root. The squaring stage is what makes the measure sensitive to outliers: a single return three times further from the mean than another contributes nine times as much to the sum.

Because the final square root restores the original units, a monthly return series measured in percent yields a standard deviation in percent, which is why practitioners quote volatility rather than variance in client reporting even though variance is the quantity that aggregates cleanly in portfolio mathematics.

Variance, not standard deviation, is the additive building block. For two assets, portfolio variance equals the first weight squared times the first variance, plus the second weight squared times the second variance, plus twice the product of the two weights and the covariance between the assets. Covariance can be rewritten as the correlation coefficient multiplied by both standard deviations, which makes the diversification mechanism explicit: whenever correlation is below one, the cross term is smaller than it would otherwise be and the square root of the resulting variance falls below the weighted average of the component volatilities.

Only at perfect positive correlation do the two coincide. This single relationship underpins mean-variance optimisation, the efficient frontier and the distinction between diversifiable and systematic risk.

Standard deviation sits in the denominator of the Sharpe ratio, converting excess return into return per unit of total risk, and in the coefficient of variation, which scales dispersion by the mean to compare series of different magnitudes. Parametric value at risk multiplies portfolio volatility by a z-score drawn from the normal distribution. Tracking error is the standard deviation of active returns against a benchmark, and investment mandates frequently cap it. Option pricing inverts the logic entirely: given a market price, the Black-Scholes-Merton framework solves for the implied volatility consistent with it, and indices such as the VIX aggregate those implied figures across strikes.

Performance presentation standards commonly require a trailing three-year annualised standard deviation for composites and their benchmarks.

A candidate should be able to compute both sample and population versions from raw data, know that financial series almost always call for the sample formula, convert between variance and standard deviation in either direction, and apply the square-root-of-time scaling to move between frequencies. Under a normality assumption, roughly two thirds of observations fall within one standard deviation of the mean, about ninety-five percent within two and almost all within three; when normality cannot be assumed, Chebyshev's inequality gives weaker but distribution-free bounds.

The limitations matter as much as the mechanics. The measure treats upside and downside dispersion identically, which motivates target semideviation and downside risk measures, and it understates tail risk in return distributions that are negatively skewed or leptokurtic, as most equity and credit series are.