Stock volatility measures variation in a share's returns; historical and implied estimates differ, and neither predicts direction or sets a maximum loss.
Stock volatility measures how widely a stock’s returns vary over a specified period. A common measure is the standard deviation of periodic percentage returns, often converted to an annualized figure.
Volatility describes dispersion, not whether a stock is rising or falling. A large positive move and a large negative move can both increase it. It is useful for comparing return variability and understanding option pricing, but it is not a complete measure of investment risk.
| Measure | Inputs | What it describes | Main limitation |
|---|---|---|---|
| Historical volatility | Past returns over a chosen lookback window | Variability observed in that sample | A different sample or future period can behave differently. |
| Implied volatility | Current option prices and an option-pricing model | The volatility input consistent with those prices and model assumptions | It varies with the option and is not a guaranteed forecast. |
The Options Industry Council’s technical explanation of volatility distinguishes historical observations from estimates inferred from options. It also explains why options with different strikes or expirations can have different implied volatilities.
A stock’s implied volatility is therefore not necessarily one unique number. A data service might display an at-the-money estimate, an average across options, or a constant-maturity calculation. Check which measure is shown before comparing it with historical volatility.
Start with returns, rather than the dollar share-price levels. For split-consistent prices, a simple price return for one period is:
Pt is the ending price and Pt-1 is the preceding price. For example, $50 to $51 produces a 2% price return.
If the purpose is to measure total-return volatility, include dividends consistently rather than silently mixing price returns and total returns. Some calculations use logarithmic returns instead of simple returns; the method should be stated and kept consistent.
The sample standard deviation is:
Here, ri is an individual return, the bar over r denotes the sample’s arithmetic mean, N is the number of returns, and s is volatility at the same frequency as those returns. There must be at least two observations.
The N-1 denominator makes this a sample calculation. The NIST explanation of measures of scale gives the variance and standard-deviation formulas. Taking the square root restores the original units after deviations have been squared.
Suppose a hypothetical stock has these ten daily returns:
2%, 3%, -1%, 4%, 2%, -3%, 5%, -2%, 1%, 3%.
For clarity, the arithmetic below uses percentage-point numbers such as 2 and -1, rather than decimal returns such as 0.02 and -0.01.
| Step | Calculation | Result |
|---|---|---|
| Mean daily return | (2 + 3 - 1 + 4 + 2 - 3 + 5 - 2 + 1 + 3) / 10 | 1.4% |
| Sum of squared deviations | (2 - 1.4)^2 + (3 - 1.4)^2 + … + (3 - 1.4)^2 | 62.4 squared percentage points |
| Sample variance | 62.4 / (10 - 1) | 6.9333 squared percentage points |
| Daily standard deviation | Square root of 6.9333 | Approximately 2.63% |
The 1.4% mean describes average daily return in this sample; the 2.63% standard deviation describes variation around that mean. They are not interchangeable.
Using decimal inputs gives the same result: the daily standard deviation is approximately 0.02633, expressed as 2.63%. Keep one unit convention throughout the calculation.
A common comparison convention scales periodic standard deviation by the square root of the number of periods in a year:
For this example, assume 252 trading days per year. Using the unrounded daily result gives:
This scaling assumes stable periodic variance and uncorrelated returns across periods. It follows the variance-addition rule under those assumptions; it is not an exact calculation of the standard deviation of compounded annual simple returns.
Ten days are a very short sample. Reporting 41.80% does not make the estimate a reliable forecast of the next year. It also does not mean a 41.80% loss is expected or that larger losses are impossible.
Keep the lookback window separate from the reporting scale: a 20-day historical estimate can be annualized without having observed a full year.
A stock’s standard deviation measures its own return variability. Beta instead measures return sensitivity to a specified market benchmark. A low beta does not rule out large company-specific price swings.
VIX is different again. Cboe describes it as an annualized, non-directional measure of expected S&P 500 volatility over a 30-day horizon, derived from SPX option prices. It is not the volatility of an individual company. See Cboe’s explanation of what VIX measures.
Neither a stock’s historical volatility nor VIX tells an investor what return to expect from that stock.
Volatility is one input to analysis, not a stand-alone buy, sell, or position-sizing rule. Stocks and options can produce substantial losses. This explanation is educational, not personalized investment advice.