Standard Deviation

March 28, 2026 Risk Management

Standard deviation measures how widely returns vary around their average and is commonly used as a historical volatility measure.

On this page

Standard deviation measures how widely observations vary around their average. In finance, the standard deviation of periodic returns is commonly used as a measure of historical volatility: a higher value means returns were more dispersed during the selected period.

Standard deviation measures variability, not investment quality or total financial risk. It treats gains and losses symmetrically and does not show liquidity, maximum loss, default, drawdown, or the order in which returns occurred.

Key Takeaways

Standard deviation is the square root of variance and is expressed in the same units as the observations.
The result depends on return frequency, lookback period, data source, currency, fees, and sample-versus-population convention.
Historical standard deviation describes the selected history; implied or forecast volatility uses a model or market prices.
Annualization by a square-root rule requires assumptions that may fail for serially correlated, illiquid, or regime-changing returns.
A normal-distribution interpretation is valid only when the distribution assumption is reasonable.
Downside and drawdown measures add information standard deviation does not capture.

Sample Standard Deviation Formula

For a sample of \(n\) returns:

s = \sqrt{ \frac{ \sum_{i=1}^{n}(R_i-\bar{R})^2 }{ n-1 } }

where:

\(R_i\) is the return in period \(i\)
\(\bar{R}\) is the sample average return
\(n-1\) is the sample degrees-of-freedom denominator

If the observations are treated as the entire population of interest, a population calculation may divide by \(n\) instead. Financial data systems should state which convention they use.

Worked Example

Assume five monthly returns:

2%, -1%, 4%, 0%, and 5%

The average is 2%. Deviations from the mean are:

0, -3, 2, -2, and 3 percentage points

The squared deviations sum to 26. Sample standard deviation is:

s = \sqrt{\frac{26}{5-1}} = \sqrt{6.5} \approx 2.55\%

Using the population denominator would produce approximately 2.28%. The same return history therefore has two valid results for different statistical purposes.

Interpreting Standard Deviation

A monthly standard deviation of 2.55% means monthly returns in the selected sample were dispersed around their average by that statistical scale. It does not mean:

the investment normally loses 2.55%
losses cannot exceed 2.55%
the next month will fall within 2.55% of the average
the return distribution is normal
a higher-volatility investment will have a higher return

If returns were independent and normally distributed, about 68% of observations would be expected within one standard deviation of the mean. Real financial returns can be skewed, fat-tailed, serially dependent, and subject to jumps, so that rule should not be assumed automatically.

Historical, Implied, and Forecast Volatility

Measure	Main input	What it represents
Historical standard deviation	Realized past returns	Dispersion during the selected history
Rolling volatility	Moving historical window	How measured dispersion changes over time
Implied volatility	Option prices and a pricing model	Volatility input consistent with observed option prices
Forecast volatility	Statistical or risk model	Model estimate for a future horizon

These measures are not interchangeable. Implied volatility also depends on option pricing assumptions and supply and demand, while forecast volatility depends on model specification.

Annualizing Standard Deviation

Periodic standard deviation is often annualized:

\sigma_{\text{annual}} \approx \sigma_{\text{periodic}} \sqrt{m}

where \(m\) is the number of periods per year, such as 12 for monthly data or a selected trading-day convention for daily data.

The shortcut assumes stable variance and sufficiently independent periodic returns. It can mislead when:

volatility clusters
returns are autocorrelated
prices are stale or smoothed
leverage changes
options create nonlinear returns
the portfolio is rebalanced
markets close or become illiquid

State the annualization convention rather than comparing an annual figure with a monthly one.

Return Measurement Choices

Standard deviation changes with:

arithmetic versus logarithmic returns
daily, weekly, monthly, or annual observations
total return versus price return
gross versus net-of-fee performance
base currency and foreign-exchange conversion
treatment of distributions and capital flows
missing values and non-trading days
overlapping versus non-overlapping periods

For funds, external subscriptions and withdrawals should not be mistaken for investment return. For illiquid assets, appraisals and stale prices may artificially smooth measured volatility.

Portfolio Standard Deviation

Portfolio volatility depends on each position’s volatility and how returns move together. For a two-asset portfolio:

\sigma_p^2 = w_1^2\sigma_1^2 + w_2^2\sigma_2^2 + 2w_1w_2\sigma_1\sigma_2\rho_{12}

where \(w\) values are portfolio weights and \(\rho_{12}\) is correlation.

Diversification can reduce portfolio volatility when assets are not perfectly positively correlated. The benefit is model- and period-dependent; correlations can increase during stress.

Standard Deviation vs. Downside Measures

Measure	Main focus
Standard deviation	Dispersion above and below the mean
Semivariance	Squared deviations below a mean or target
Downside deviation	Square root of target semivariance
Maximum drawdown	Largest observed peak-to-trough decline
Ulcer Index	Root-mean-square historical drawdown
VaR	Loss quantile at a stated horizon and confidence level
Expected shortfall	Average modeled loss in the selected tail

A strategy can have low standard deviation and severe tail risk if it earns many small gains and occasional large losses.

How to Compare Investments

Align:

return period and lookback window
currency
fees and distributions
sampling frequency
annualization convention
benchmark and market regime
data quality and survivorship

Then review return, drawdown, liquidity, concentration, leverage, and downside measures rather than ranking solely by standard deviation.

Risks and Limitations

Symmetry: gains increase measured risk just as losses do.
Historical dependence: the selected window may be unusually calm or volatile.
Tail behavior: one scale statistic does not describe skew, jumps, or fat tails.
Path blindness: return order and recovery time are ignored.
Smoothing: stale or model-based prices can understate variation.
Instability: volatility and correlation can change rapidly.
No valuation conclusion: low volatility does not mean an asset is fairly priced or safe.

Common Mistakes

Calling standard deviation the average return difference.
Mixing sample and population formulas.
Comparing daily and annualized values.
Applying square-root annualization without checking assumptions.
Assuming returns are normal because standard deviation is reported.
Treating low historical volatility as guaranteed stability.
Ignoring fees, currency, and total-return adjustments.
Using standard deviation as the only measure of risk.

Authoritative Context

Downside Risk: Focuses on outcomes below a defined threshold rather than treating gains and losses symmetrically.
Semivariance: Measures squared shortfalls below a selected mean or target.
Beta: Measures sensitivity to benchmark returns rather than total stand-alone dispersion.
Value at Risk: Estimates a loss threshold at a stated confidence level and horizon.
Tail Risk: Concerns unusually severe outcomes that standard deviation alone may not describe well.

FAQs

What does standard deviation mean for an investment?

It describes how widely the investment’s periodic returns varied around their average during the selected period.

Is higher standard deviation always worse?

No. It means greater dispersion, not automatically a worse investment. Expected return, downside, liquidity, horizon, and investor circumstances also matter.

Does one standard deviation always contain 68% of returns?

Only under a normal-distribution assumption. Actual financial returns may not follow that distribution.

Educational Use

This article provides general financial education. It is not personalized investment, trading, portfolio-construction, statistical, tax, legal, or risk-management advice.

Semivariance

Ulcer Index