Direct Answer
Rolling volatility is a volatility measure computed as the standard deviation of an asset's returns over a moving (rolling) lookback window, recalculated each period as the window shifts forward. Each day the calculation drops the oldest return from the window and adds the newest one, so the figure updates continuously rather than being fixed to a single historical period.
Key Takeaways
- Rolling volatility is the standard deviation of an asset's daily returns over a moving lookback window, commonly 20 or 30 days, recalculated each period as the window shifts forward.
- Annualized volatility multiplies the daily rolling volatility figure by the square root of 252, the approximate number of trading days in a year, for daily data.
- Rolling volatility rises and falls as calm or turbulent periods enter and exit the lookback window; it says nothing about direction, only magnitude of movement.
- It's backward-looking and distinct from implied volatility, which is derived from current option prices and reflects a forward-looking expectation.
- A shorter window reacts faster to fresh turbulence; a longer window smooths noise but adjusts more slowly to a genuine regime change.
What Is Rolling Volatility?
Rolling volatility is a volatility measure computed as the standard deviation of an asset's returns over a moving (rolling) lookback window, recalculated each period as the window shifts forward. Each day the calculation drops the oldest return from the window and adds the newest one, so the figure updates continuously rather than being fixed to a single historical period. Because it's built from standard deviation, rolling volatility captures the size of an asset's day-to-day swings without regard to whether those swings were up or down, a sharp rally and a sharp selloff of similar magnitude can produce a similar rolling volatility reading.
The Formula
Rolling Volatility = standard deviation of daily returns over the last n days (commonly 20 or 30 days).
Annualized Volatility = Rolling Volatility × √252 for daily data.
The lookback window, n, sets how many of the most recent daily returns feed the standard-deviation calculation. Because the standard deviation of daily returns is a small daily figure, it's often annualized so it can be compared across assets or against conventions used elsewhere in markets, such as options pricing. Multiplying by the square root of 252, the approximate number of trading days in a calendar year, converts a daily standard deviation into an annualized one, under the assumption that returns are being sampled daily.
Worked Example
Hypothetical example, for education only.
Suppose a stock's daily returns over the last 20 trading days have a standard deviation of 1.80%. That figure, 1.80%, is the 20-day rolling volatility for that day.
Annualized volatility = 1.80% × √252 ≈ 1.80% × 15.87 ≈ 28.6%.
Now suppose the market enters a turbulent stretch and, a few weeks later, the 20-day standard deviation of daily returns has risen to 3.20%. Annualized volatility = 3.20% × √252 ≈ 50.8%. Nothing about the underlying formula changed, only the dispersion of the returns inside the 20-day window did. As the calm days that preceded the turbulent stretch roll out of the window and are replaced by the more volatile recent days, the rolling volatility reading climbs; once the turbulence itself ages out of the window, the reading falls back down, even if the market hasn't yet calmed by any other measure.
How Rolling Volatility Is Used
Reading the trend of the reading itself
Traders commonly track whether rolling volatility is rising, falling, or flat rather than focusing on any single day's value. A rising reading indicates that recent returns have been more dispersed than the prior stretch inside the window; a falling reading indicates the opposite. Because the window moves, a spike in volatility eventually rolls out and the reading recedes even without any new information, the level reflects what's currently inside the window, not a permanent state.
Sizing positions and setting stops
Some position-sizing and stop-placement approaches scale with recent volatility, using a wider stop or a smaller position size when rolling volatility is elevated, and the reverse when it's low, the goal being to hold risk roughly constant in dollar terms as an asset's typical daily movement changes.
Comparing regimes and assets
Annualized rolling volatility puts assets and time periods on a common, comparable scale. A stock currently showing 50% annualized volatility is moving, on a standard-deviation basis, roughly twice as much day to day as one showing 25%, useful context whether comparing two different assets or the same asset in a calmer period versus a turbulent one.
A baseline for option and derivatives pricing
Rolling (historical) volatility is one input traders reference when evaluating whether an option's implied volatility looks rich or cheap relative to how the underlying asset has actually been trading recently, though implied volatility is a forward-looking market expectation, not a direct forecast derived from the historical figure.
Common Lookback Windows
| Window (n days) | Responsiveness | Common use |
|---|---|---|
| 10 | Fastest, noisiest | Short-term/tactical monitoring |
| 20 | Balanced (roughly one trading month) | General-purpose volatility tracking |
| 30 | Slightly smoother than 20-day | General-purpose volatility tracking |
| 60 | Slower, smoother | Longer-term regime comparison |
20 and 30 days are the two most commonly cited windows for rolling volatility, but the choice is a convention rather than a fixed rule, verify the default your specific charting platform or data provider uses, since it can vary.
Limitations
- Backward-looking by construction, every input is a past daily return, so rolling volatility describes what already happened inside the window, not what will happen next.
- Window-length sensitivity, a shorter window can show a sharp spike that a longer window barely registers, and vice versa; the same asset on the same day can display noticeably different rolling volatility readings depending on n.
- One large return can dominate a short window, a single outsized daily move sits inside a 20-day window for a full month, elevating the reading even after conditions have otherwise normalized, until that one day finally rolls out.
- No directional information, a large up day and a large down day contribute similarly to standard deviation, so rolling volatility alone can't distinguish a rally from a selloff.
- The annualization assumes daily sampling and roughly stable behavior across the year, multiplying by the square root of 252 converts units, but it doesn't correct for the fact that volatility itself is not constant across the year it's being annualized to.
Common Mistakes
- Comparing rolling volatility readings computed with different window lengths as if they were the same measure, a 10-day and a 60-day reading on the same asset can look meaningfully different and both be correct for their own window.
- Treating rolling (historical) volatility and implied volatility as interchangeable, one looks backward at realized returns, the other looks forward at option-market pricing, and they can diverge.
- Reading a single elevated day as a lasting regime change without checking whether it reflects one outlier return still sitting inside the window.
- Forgetting to confirm annualization assumptions, the √252 convention assumes daily data; applying it to weekly or monthly returns without adjusting the trading-period count produces an incorrect annualized figure.
Why the Reading Is Lowest Just Before It Matters
Rolling volatility computes dispersion over a trailing window, which means a long quiet stretch produces a low reading regardless of what conditions are building. The measure is at its most reassuring precisely when the data it has seen is least representative of what follows.
That property argues for using it as a sizing input rather than a risk assessment. Scaling positions inversely to measured volatility keeps risk contribution roughly stable through changing conditions, and it does so without requiring the measure to predict anything. What it cannot do is warn you.
The mistake is comparing readings across different window lengths or annualisation conventions. A figure computed over twenty periods and one over sixty describe different things, and annualised figures depend on an assumption about how volatility scales with time that holds only approximately.
The measure also treats upside and downside movement identically. An asset that rose sharply and one that fell sharply produce the same reading, which is appropriate for some purposes and misleading if the number is being read as a measure of danger.
Rolling Volatility FAQs
What is rolling volatility?
Rolling volatility is the standard deviation of an asset's returns calculated over a moving lookback window, commonly the last 20 or 30 days, and recalculated each period as the window shifts forward one day at a time.
How is rolling volatility calculated?
Take the daily returns over the last n days (commonly 20 or 30), compute their standard deviation, and that figure is the rolling volatility for that day. The next day, the window drops the oldest return and adds the newest one, and the calculation repeats.
Why is rolling volatility annualized?
A daily standard deviation is hard to compare across assets or to option-pricing conventions, which are typically quoted on an annual basis. Multiplying by the square root of 252, the approximate number of trading days in a year, converts a daily figure into an annualized one, assuming daily data.
What lookback window is commonly used for rolling volatility?
20 and 30 trading days are the most commonly cited windows, roughly matching one and one-and-a-half calendar months. Shorter windows react faster to new turbulence; longer windows smooth out day-to-day noise but adjust more slowly.
Is rolling volatility the same as implied volatility?
No. Rolling volatility is backward-looking, computed directly from an asset's own past returns. Implied volatility is forward-looking, backed out from current option prices and reflecting the market's expectation of future volatility, the two can diverge, sometimes sharply.
Does higher rolling volatility mean higher risk?
Higher rolling volatility means larger recent price swings in either direction, which is one common way to define risk, but it says nothing about direction, a sharp rally and a sharp selloff can produce the same rolling volatility reading.
Why do rolling volatility readings drop sharply when a large move leaves the window?
Each observation enters and exits the window with equal weight, so a single extreme day continues to inflate the reading until it drops out, at which point the value falls abruptly. The drop reflects the window's composition changing rather than any change in market conditions that day. Exponentially weighted alternatives avoid this artifact by giving older observations progressively less weight.
Should volatility be computed on simple or logarithmic returns?
Logarithmic returns are conventional because they are additive across periods, which makes the annualisation arithmetic consistent. The two produce nearly identical results for small moves and diverge for large ones, so the choice matters most for volatile instruments. What matters more than the choice is applying the same convention across everything being compared.
Does higher measured volatility mean higher expected return?
Realised volatility describes how much prices moved, not what returns followed. The relationship between the two has been studied extensively and is not the simple positive link that risk-and-return framing implies, particularly over shorter horizons. Using measured volatility as a sizing input is well founded; using it as a return forecast is not.