Comparing realized-volatility estimators
Compare closing-return and range-based methods by the price information each uses and the assumptions each needs.
By OptionsKing. Publisher and editorial methods.
What you will learn
By the end of this lesson, you should be able to:
- Identify estimator inputs.
- Separate overnight and intraday movement.
- Calculate a stated variance combination.
A volatility estimator is a rule for turning price observations into a measure of variability. Different rules use different information from the same trading sessions. Comparing them is useful only when the data window, adjustments, and annualization convention are kept clear.
Compare the inputs
Close-to-close standard deviation uses successive closing returns. Parkinson uses session highs and lows. The basic Garman-Klass form uses the open, high, low, and close under assumptions about the price process and opening jumps. These methods need not give the same result.
Yang-Zhang combines an overnight-return variance, an open-to-close return variance, and an intraday range component based on Rogers-Satchell. The latter uses high and low prices relative to the open and close to estimate session variance while allowing drift.
Locate the overnight move
Suppose a fictional stock closes at $100, opens the next session at $90, trades between $88 and $92, and closes at $90. The overnight log return is ln(90 / 100), approximately -10.54%. The open-to-close return is zero.
The close-to-close return also includes the net -10.54% change. It is not blind to the gap, but it does not separately identify the overnight and session contributions. A high-low range alone describes only the $88 to $92 session and omits the direct $100-to-$90 opening move.
Combine stated component estimates
For Yang-Zhang, daily variance is overnight variance plus k times open-to-close variance plus (1 - k) times the Rogers-Satchell component. A common weight is k = 0.34 / [1.34 + (n + 1) / (n - 1)] for n observations greater than one.
For 20 observations, k is approximately 0.139. Suppose the three component estimates are 0.0004, 0.0001, and 0.0002 respectively. The combination is about 0.0005861. Annualizing with 252 sessions and taking the square root gives approximately 38.43% volatility.
Those component values are a separate illustrative input set, not estimates from the single price bar above. Multiple observations are required to estimate the variances.
Choose a consistent measurement
Range estimators can use more information per session, but their properties depend on assumptions and data quality. Bad highs or lows, inconsistent price adjustments, and omitted sessions can distort the estimate.
State the method when comparing historical volatility with IV. A gap of a few volatility points can change with the estimator or window. No historical estimator eliminates uncertainty about the next period or provides a guaranteed trading signal.
Check your understanding
Why can estimators differ on the same sessions?
Let's walk through the answer.
The rule determines which observed movements enter the calculation.
Sources
Rules and thresholds above were checked against these documents on August 9, 2026. Exchange and broker rules change. Confirm anything you are about to act on with your own broker.
Keep reading
Do the math on your own trade
Every price in this article is an illustrative worked example, not a quote. Read Volatility for the rest of the series, and the disclaimer before you act on any of it. Selling options carries real risk of loss, and the loss can be far larger than the premium you collected.