Skip to main content

Volatility

Volatility measures how returns vary around a reference value but does not capture every form of investment risk.

Published
Updated
Reviewed
Difficulty
Beginner
Reviewer
JH

Overview

Volatility describes how widely and how often returns move around a reference value. It helps quantify variability, but it does not explain every way an investment can lose money.

The number depends on the data frequency, time window, return definition, and calculation method. Historical volatility describes observed fluctuations; it does not determine future behavior.

[1]

Return dispersion becomes a volatility estimate

Suppose an investment has periodic returns of 1%, 1%, 1%, and 1%. Its return is not certain in any broader sense, but the observed dispersion in this sample is zero. Another sequence of −5%, 7%, −4%, and 6% has a similar average yet much wider movements and therefore higher measured volatility.

A common historical estimate uses standard deviation:

σ=i=1n(rir¯)2n1

where

σ
the sample standard deviation of returns
ri
the return in observation i
r¯
the arithmetic mean return in the sample
n
the number of return observations

The calculation finds how far each return lies from the sample average, squares those differences so negative and positive deviations do not cancel, averages them with a sample adjustment, and takes the square root to return to the original unit. A result of 12% means dispersion is measured in percentage-return terms; it does not mean the investment will stay within 12% of its average.

Different formulas may use population rather than sample standard deviation, log returns rather than simple returns, or model-based weights. The method should be consistent when figures are compared.

Frequency and time window can change the answer

Daily, weekly, monthly, and annual returns contain different information. A volatility estimate based on five quiet years may be much lower than one based on a short period containing a market shock. Neither is automatically wrong; they summarize different samples.

Short-interval volatility is often annualized by multiplying by the square root of the assumed number of periods in a year. For example, monthly volatility may be multiplied by the square root of 12.

That scaling relies on strong assumptions, including that returns are sufficiently independent and their distribution is stable through time. Financial returns can cluster into calm and turbulent periods, so annualized volatility can give a misleading sense of precision when those assumptions fail.

Historical, implied, and expected volatility

Comparison criterionHistorical volatilityImplied volatility
Based onObserved past returnsCurrent option prices and a pricing model
DescribesRealized past variationThe volatility input consistent with option prices
Main limitationThe future may differ from the sampleDepends on model assumptions and market pricing

Expected volatility is a forward-looking estimate from a model, forecast, or market view. Implied volatility is also forward-looking in interpretation, but it is not a direct consensus forecast of actual future standard deviation. Option prices also reflect supply, demand, risk preferences, and the assumptions of the model used to infer the number.

Terms such as "a volatility of 20%" are incomplete unless the measurement basis is understood. The asset, horizon, return interval, method, and whether the figure is historical or implied all matter.

Volatility is not the same as risk

Volatility treats upside and downside deviations alike. A sharp gain increases volatility even though investors generally do not regard the gain itself as a loss. Conversely, a position can appear stable for a long time and still contain the possibility of a sudden default, trading halt, or illiquidity event.

Risks that volatility may not capture well include:

  • Permanent loss: an issuer can fail or an asset can become impaired.
  • Liquidity risk: quoted prices may not be achievable for a large or urgent sale.
  • Concentration risk: one event can affect a large share of a portfolio.
  • Path and timing risk: a loss just before a required withdrawal can matter more than the same average volatility at another time.
  • Model risk: a short or unusually calm dataset can understate the range of possible outcomes.

How portfolio holdings affect each other

Portfolio volatility depends on the volatility of each holding and on how their returns move together. Combining two volatile assets can reduce total volatility when their returns are not perfectly aligned. If they tend to fall and rise together, the reduction may be small.

This relationship is summarized through correlation or covariance. Those estimates can change, especially during market stress. Diversification can reduce some variability, but it cannot guarantee a maximum loss or eliminate risks shared across holdings.

Frequently Asked Questions

Continue learning

Sources

  1. [1]
    Risk

    Investor.gov, U.S. Securities and Exchange Commission