Direct answer
ATR typically “behaves differently” when the underlying volatility conditions change, because ATR measures the average size of recent price movement. It tends to increase during periods with larger true ranges (more volatility, wider swings, or gap-like moves), and decrease when true ranges contract (quieter, more range-bound trading). At the same time, ATR can also look different because its computation depends on the chosen lookback period and on how price transitions occur (for example, gaps versus smooth movement).
This article focuses on conditional behaviour (what ATR will tend to reflect under different conditions), not on predicting future price movements or promising performance.
Mechanism or definition
ATR stands for Average True Range. The “true range” for a period is based on the size of price movement and includes the effect of gaps relative to the prior close. ATR is then computed as an average of those true ranges over a chosen number of periods.
The key point is conditional behaviour: ATR is not a fixed property of a market. It is a statistic built from recent observations. If recent true ranges are systematically larger, the average will rise; if they are systematically smaller, the average will fall.
Two variables matter for interpretation:
- Market condition (inputs): how large the true range observations are in recent periods.
- Calculation condition (settings): the lookback window and the timeframe used to compute the true ranges.
Evidence or example (factual comparison)
Consider the same instrument computed with two common settings: a short lookback (more responsive) and a longer lookback (more stable).
1) Volatility regime shift (expansion vs contraction)
- In a volatility expansion phase, candles often show wider intraperiod swings and/or larger gap distances between closes and opens. Because those periods contribute larger true ranges, the ATR average increases.
- In a volatility contraction phase, the average true range observations are smaller, so the ATR average decreases.
The observable “difference” is that ATR may climb quickly in expansion and then decay more slowly when expansion stops, depending on the length of the lookback window.
2) Timeframe changes (adaptation speed)
Compute ATR on a faster timeframe versus a slower one:
- On a short timeframe, true ranges are measured over shorter intervals, so ATR can change more quickly when the market’s short-term behaviour shifts.
- On a long timeframe, ATR aggregates more history per point. Even if short-term volatility spikes occur, the longer averaging window can dampen how quickly ATR moves.
So the same underlying market can show different ATR “behaviour” purely because of the timeframe and averaging window.
3) Non-smooth price transitions (gaps and discontinuities)
ATR’s true range calculation is designed to account for discontinuities between periods (for example, when an open is far from the prior close). When markets experience more abrupt transitions, those periods can increase true range values, and therefore pull ATR upward more than smooth, incremental movement would.
4) Costs and execution effects (not directly in ATR)
ATR is computed from price observations, not from trading costs. Spreads, commissions, slippage, and market liquidity can vary when volatility changes. In practice, the distance implied by “typical movement” (as summarized by ATR) may not match what is achievable after costs. This creates a conditional mismatch: ATR can rise due to volatility, while trading frictions also rise, making outcomes diverge from any simplistic expectation.
Limitations and risks
- No forecasting guarantee: ATR is a descriptive statistic of recent movement. Historical relationships between ATR and outcomes do not ensure future results.
- Sensitivity to settings: Different lookback periods and timeframes can produce noticeably different ATR paths for the same market, especially during transitions.
- Provider/platform calculation details: ATR output can vary if inputs differ (such as how price data is formed, which trading hours are included, or how true range is computed by a specific platform). Without matching calculation assumptions, comparisons can be unreliable.
- Failure mode in changing dynamics: ATR may respond to volatility changes, but it does not identify whether the change is due to structural shifts, temporary shocks, scheduled events, or changes in market microstructure.
- Cost/implementation gap: ATR does not include transaction costs or execution quality, which can change materially when volatility changes.