What Realized Volatility Is
Realized volatility is a historical measure of how widely an asset’s returns have dispersed over a completed observation window. It is built from recorded price changes rather than from expectations about what may happen next, so it belongs entirely to the historical record. Once the measurement window ends, the underlying inputs are fixed, and the reading becomes a summary of variability that has already occurred.
At its core, realized volatility tracks dispersion in returns rather than direction in price. A market can rise steadily and still show low realized volatility if the path is orderly, while a market that goes nowhere overall can show high realized volatility if returns swing sharply from one period to the next. That distinction is what makes realized volatility a measured form of volatility rather than a synonym for trend, momentum, or market sentiment.
How Realized Volatility Is Observed
Realized volatility is formed from a sequence of actual price changes across a defined lookback window. It does not come from a single move or a single endpoint. Instead, it reflects the internal path of returns across the period, which is why a sharp jump followed by calm trading produces a different reading than repeated back-and-forth movement of similar net size.
Mechanically, the measure starts with a return series sampled over a chosen interval, such as daily close-to-close returns or intraday price changes. Those returns are then aggregated into a dispersion measure, most often realized variance or its square-root form, realized volatility. In practical terms, the calculation asks how spread out the returns were inside the window rather than whether the asset ended that window up or down. A string of similar small moves produces a lower reading than a sequence of uneven returns with the same net result.
The return series matters because realized volatility is only as specific as the observations used to build it. A close-to-close series captures one completed move per session, while an intraday series captures more of the path inside the session. Arithmetic returns and log returns usually point in the same broad direction for interpretation, but they are still different constructions of the historical path. For that reason, realized volatility should be understood as a measure built from observed returns, not as a generic number detached from how those returns were defined.
Core Inputs That Shape the Reading
The first structural input is the lookback window. A 10-day realized volatility reading and a 60-day realized volatility reading are not competing descriptions of the same object so much as summaries of different slices of history. Shorter windows react faster to bursts, reversals, and volatility clustering because recent observations carry more weight in the sample. Longer windows smooth those changes by blending current movement into a broader record of past returns.
The second input is sampling frequency. Intraday data can capture more short-term variation than daily data, and daily data captures more internal texture than weekly data. That means two readings can differ materially even when they are built over the same calendar period, because one method observes more of the price path than the other. Realized volatility therefore depends not only on how long the window is, but also on how finely the return series is sampled inside that window.
The third input is horizon presentation. Realized volatility is often annualized so that readings from different windows can be expressed on a common scale, but annualization is a scaling convention rather than a new source of information. It makes a short-horizon estimate easier to compare with other standardized volatility numbers, yet the underlying reading still describes the completed horizon from which it was derived. In that sense, realized volatility is always horizon-dependent first and standardized only second.
Common Forms of Realized Volatility
Realized volatility can be classified by how returns are sampled and by what exactly is being measured. A close-to-close version uses one return per day and offers a simple summary of completed daily movement. An intraday or high-frequency version uses many observations inside the trading day and can capture more of the path that daily closing prices leave out. A rolling version updates the window continuously, while a fixed-window reading describes one completed period without refreshing the sample each day.
The measure can also be applied at different levels. It may describe a single asset, a sector basket, an index, or even a portfolio if the underlying returns are combined appropriately. That is why realized volatility is best understood not as one rigid formula but as a family of historical dispersion measures built from observed returns. The unifying feature is always the same: the reading summarizes variability that has already been recorded in market prices.
What Changes Realized Volatility
Realized volatility rises when price moves become larger, less even, and more clustered. Repeated wide swings, abrupt reversals, gaps, and unstable intraperiod movement all push the reading higher because they widen the dispersion of returns inside the measurement window. One shock can lift the measure for a time, but a series of large moves tends to keep it elevated for longer because fresh dispersion keeps entering the sample.
It falls when price changes stay relatively small or when the market moves in a smoother, more orderly way. That is why low realized volatility does not necessarily mean an inactive market. Price can still travel a meaningful distance if the move is steady and interruptions are limited. What matters is not whether the asset ended higher or lower, but how unevenly it traveled through time.
Where Realized Volatility Fits in the Framework
Within the broader volatility and stress framework, realized volatility sits on the observed side of market variability. It is narrower than volatility in general because it refers specifically to movement that has already been recorded in the price path. That makes it a descriptive measure of past instability rather than a catch-all label for turbulence.
Its closest neighboring concept is implied volatility. Realized volatility describes variability that has already happened, while implied volatility reflects expected variability embedded in forward-looking pricing. They belong to the same family, but they point in different informational directions.
The distinction also matters in discussions of the volatility risk premium. That premium depends on the relationship between expected and observed volatility, and realized volatility provides the observed side of that relationship. In that sense, it serves as a recorded benchmark for comparison rather than a forward-looking pricing signal.
What Realized Volatility Does Not Tell You
Realized volatility does not tell you whether the market was moving up or down. It does not tell you why volatility increased, and it does not identify a regime on its own. A high reading may coincide with macro shocks, liquidity stress, earnings events, forced repositioning, or broader instability, but the number itself does not separate those causes. It compresses many possible drivers into one descriptive measure of dispersion.
It also does not function as a standalone forecast. Because it is built from completed observations, it cannot by itself determine what volatility will do next. It can help describe recent conditions and provide context for broader interpretation, but it should not be treated as an independent rule for expectation, classification, or action. Its value is strongest when it is used as a historical measurement input rather than a self-sufficient market conclusion.
FAQ
Can realized volatility be high in a rising market?
Yes. Realized volatility measures the size and unevenness of price changes, not whether the asset finished higher or lower. A rising market with sharp pullbacks and wide daily ranges can produce a high reading, while a smooth uptrend can produce a lower one.
Why can two realized volatility readings for the same asset look different?
They may use different lookback windows, different return series, different sampling frequencies, or different scaling conventions. A short-window daily reading and a longer-window intraday-based reading are describing different slices and textures of the same price history, and annualized versions are still built from those underlying horizon choices.
Is realized volatility the same thing as implied volatility?
No. Realized volatility is backward-looking because it is built from completed price moves. Implied volatility is forward-looking in the sense that it is inferred from current option pricing. They are related, but they are not interchangeable.
Does one big shock keep realized volatility high forever?
No. A large move can raise the reading sharply, but its influence fades as newer observations replace it in the measurement window. Realized volatility tends to stay high for longer when large moves keep arriving in clusters rather than appearing as a one-off event.