What Atr is, in one bounded definition
Atr stands for Average True Range. It is a volatility indicator that estimates the average magnitude of recent price movement over a chosen window. The starting point is the True Range (TR), which uses the current high–low range and adjusts for gaps by also considering distances from the previous close. Atr then applies an averaging method over TR values, producing a smoother “typical” movement size than raw day-to-day ranges.
Because Atr is built from price ranges (not from returns directly), it is best thought of as an average measure of movement magnitude, not a measure of direction.
Adjacent concept 1: ATR vs “historical volatility”
Historical volatility (often abbreviated HV) typically estimates how variable returns have been over a window, using statistical dispersion of returns rather than high–low ranges. In practice, HV relies on a time series of price changes (returns), then converts that dispersion into a volatility-like scale.
Key difference:
- Atr: averages a range-based measure (True Range), emphasizing how large candles and gaps have been.
- Historical volatility: summarizes dispersion of returns, emphasizing how returns fluctuate.
Why this matters: both may rise in turbulent markets, but they can diverge when the market’s behavior changes in ways that affect ranges and gaps differently than returns dispersion. For verification, compute both from the same price series and compare how they react around events like open gaps.
Adjacent concept 2: ATR vs “range” and “rolling high–low”
A simple “range” concept measures the difference between a recent highest high and lowest low, often over the same lookback window. Some traders also track average ranges or median ranges.
Key difference:
- Atr: uses True Range at each step (including a gap adjustment), then averages those step values.
- Rolling high–low range: looks at the overall span between the window extremes, not the distribution of step-by-step movement.
Why this matters: a rolling high–low can remain elevated because it depends on window extremes; it may react slowly to the disappearance of old extremes. Atr, by contrast, can respond more smoothly because it averages the underlying step movements. A failure mode here is assuming that a “higher range” implies the same magnitude as Atr without checking the calculation method.
Adjacent concept 3: ATR vs “standard deviation of price”
Standard deviation is a dispersion statistic. Depending on implementation, it may be applied to prices directly or to returns.
Key difference:
- Atr: range-and-gap based input (True Range), then averaged.
- Standard deviation: uses deviations of a chosen series around its mean (prices or returns), then converts dispersion into a scale.
Why this matters: standard deviation on prices can be sensitive to the level of the instrument, while standard deviation on returns aligns more naturally with “volatility” as variability of change. Atr is usually designed to reflect movement size in a way tied to candle ranges. You can test these distinctions by recomputing each measure on identical data and observing whether the ranking and timing match.
Adjacent concept 4: ATR vs “directional indicators” (momentum/oscillators)
Momentum and oscillator indicators (for example, those derived from price change rates or normalized oscillations) focus on direction, speed, or relative position rather than movement magnitude.
Key difference:
- Atr: magnitude of movement (how large price swings are on average).
- Directional indicators: tendency and timing information about whether price is moving up or down.
Why this matters: you can have high Atr with weak or mixed direction (choppy movement), or low Atr during a steady trend. Treating Atr as if it encodes direction is a common conceptual mistake and a material limitation.
How Atr works mechanically (inputs, window, smoothing)
- Compute True Range (TR) each period. TR incorporates the current high–low range and includes an adjustment for gaps relative to the previous close.
- Average TR over a lookback window. The most common versions use a moving average style, producing Atr as a smoothed sequence.
- Interpret the scale as a typical movement size. Atr is expressed in the same price units as the underlying data (for example, in pips if the instrument is represented that way in the calculation environment).
Settings that change Atr
Atr values depend on the chosen lookback length and the averaging method (smoothing approach). Shorter windows generally make Atr more responsive to recent changes; longer windows typically smooth more. A material risk is comparing Atr values across platforms or configurations without matching the calculation settings, since different defaults and data handling choices can produce different numbers from the same visible chart.
Evidence and a concrete, assumption-based example
Assume you have a series of candles with known highs, lows, and previous closes. Consider two adjacent periods:
- Period A has a modest high–low range and closes near where it opened.
- Period B has a larger high–low range and/or a gap relative to the previous close.
In a True Range framework:
- TR in Period A would largely reflect the high–low size.
- TR in Period B would reflect both its high–low size and any gap distance from the previous close.
When you average TR values into Atr:
- If Period B’s TR is much larger, Atr increases, because the average includes that large movement.
- If Period B is an outlier, Atr may rise temporarily and then move back toward older typical values as newer TR observations enter the window.
This example shows the conceptual point: Atr changes when step-by-step movement (including gaps) changes, not when only extremes in a window shift.
Limitations and failure modes you can independently verify
- Not a standalone trading signal. Atr indicates typical movement magnitude; it does not specify direction, timing, or expectancy.
- Sensitive to calculation choices. Lookback length and smoothing method change the output. Always verify the exact formula and settings used by your tool.
- Sensitive to price data handling. Different data sources and time zone definitions can change candle boundaries, which affects highs/lows and gap detection.
- Market regime changes matter. Historical relationships between Atr and future outcomes do not guarantee future results, and changes in volatility clustering or liquidity can alter how “typical movement” behaves.
- Costs and execution affect what you can realize. Even if Atr helps you estimate movement magnitude, actual outcomes depend on transaction costs and execution conditions, which are not captured by a volatility statistic alone.