Realized Volatility - Measuring What Actually Happened

By EC Assets Research Team, Volatility Research · Published 13 August 2026 · Updated 13 August 2026

Realized Volatility: Realized volatility is the annualized standard deviation of an asset's returns over a lookback window: the measured record of movement that actually occurred. Every quoted figure embeds choices about window, sampling frequency and estimator, and two correct calculations on the same prices routinely disagree.

What Realized Volatility Is

Realized volatility, also called historical volatility, is the annualized standard deviation of an asset's returns measured over some past window. Where implied volatility is a price for future movement, realized volatility is the accounting of movement that already happened. The pair anchors most of professional volatility practice: implied is what the market charges, realized is what it ultimately had to deliver against.

The standard close-to-close calculation uses daily log returns:

σ = sqrt( 252 / (N-1) × Σ (rᵢ - r̄)² )

where rᵢ are daily log returns, N the number of observations, and 252 the trading days used for annualization. A daily standard deviation of 1 percent annualizes to roughly 16 percent, the arithmetic behind the rule of 16.

The Choices Hiding Inside One Number

A quoted realized volatility looks like a fact. It is the output of at least four decisions.

The window. Twenty days reacts fast and is noisy; a year is stable and stale. After a single violent week, 20-day realized doubles while 252-day barely moves, and both are correct. Any strategy keyed to realized volatility is implicitly keyed to a window choice, and the day a large move rolls out of the window, the figure drops with no change in the market whatsoever.

The mean. Subtracting the average return, or assuming it is zero, changes little over long windows and a surprising amount over short ones during strong trends. Zero-mean is the common practitioner convention, because sample means over short windows are mostly noise.

The frequency. Daily closes are the default, but volatility measured from 5-minute returns, summed as realized variance, is a far more precise estimate of the same quantity and converges much faster. Intraday measurement is what variance swap replication and modern volatility forecasting are built on, at the cost of microstructure noise if pushed too fine.

The estimator. Close-to-close throws away everything that happened between closes. Range-based estimators put the day's high and low to work: Parkinson uses the high-low range, Garman-Klass adds open and close. They achieve several times the statistical efficiency of close-to-close from the same number of days, at the price of assumptions about continuous trading and no jumps, and they miss overnight gaps entirely.

Overnight and Intraday Are Different Animals

A daily close-to-close return bundles two regimes: the overnight gap, where news accumulates while trading is closed, and the intraday session. The two components have different sizes, different dynamics and different hedgeability, and desks that manage gamma care intensely about the split, because overnight movement cannot be hedged while it happens. Equity indices routinely realize a large share of their total variance overnight, which a range-based estimator computed from session data alone will systematically miss.

Worked Example

A stock's daily log returns over 21 trading days have a standard deviation of 1.4 percent, zero-mean convention. Annualized:

σ = 1.4% × sqrt(252) ≈ 22.2%

The one-month at-the-money option was priced at an implied volatility of 27 at the start of that window. The option seller collected 27, the underlying delivered 22.2, and the difference, 4.8 volatility points, is one month's harvest of the volatility risk premium, realized after the fact. Strings of such months, punctuated by occasional months where realized prints far above implied, are exactly the return profile of systematic volatility selling.

Two weeks later, the calculation shows 34 percent: a single 6 percent earnings day entered the window. Nothing about the company's ongoing riskiness tripled; one observation dominates a short window's sum of squares. Robust practice reads the level and the composition together.

[!key] Realized volatility is the annualized standard deviation of returns over a chosen window, and every quoted figure embeds a window, a frequency, a mean convention and an estimator. A 1 percent daily standard deviation is roughly 16 percent annualized.

[!warning] Realized volatility is a lagging description, not a forecast, and it changes mechanically when large observations enter or leave the window. Strategies triggered by a realized threshold, including volatility targeting, inherit those calendar artefacts and will act on them as if they were news.

Why It Matters for Institutional Investors

• It is the settlement variable. Variance swaps and volatility swaps pay on realized, computed to a contractual formula, so the measurement conventions are not academic details but the contract itself.
• It sizes portfolios. Volatility targeting scales exposure to a realized estimate, so the window and estimator choices directly set leverage and turnover, and the mechanical drop when a crisis rolls out of the window releases risk budget with no economic cause.
• It prices the premium. The volatility risk premium is implied minus subsequently realized. Measuring realized carelessly, for instance ignoring the overnight share, silently reshapes the premium being harvested or paid.

References

1. Parkinson, M. (1980). The extreme value method for estimating the variance of the rate of return. Journal of Business, 53(1).
2. Garman, M. B., & Klass, M. J. (1980). On the estimation of security price volatilities from historical data. Journal of Business, 53(1).
3. Andersen, T. G., Bollerslev, T., Diebold, F. X., & Labys, P. (2003). Modeling and forecasting realized volatility. Econometrica, 71(2).
4. Sinclair, E. (2013). Volatility Trading (2nd ed.). Wiley.
5. Hull, J. C. (2022). Options, Futures, and Other Derivatives (11th ed.). Pearson.