Direct answer
ATR is computed from the true range over a chosen number of bars, then averaged. Because the “bars” come from a specific timeframe, the ATR value is sensitive to timeframe selection and to what holding period you are implicitly matching it to. In practice, using a shorter timeframe for the ATR calculation usually produces a more responsive estimate of recent volatility; using a longer timeframe tends to smooth volatility and respond more slowly.
Mechanics and definition
Average True Range (ATR) estimates the typical size of price movement over a selected lookback window. For each bar, the true range uses the current high and low and also references the previous bar’s close to capture gaps. ATR then averages those true ranges over N bars.
Timeframe affects ATR because it changes what a “bar” represents:
- On a higher timeframe, each bar covers a longer span of time, so the high–low movement per bar is usually larger and more aggregated.
- On a lower timeframe, each bar covers a shorter span, so the high–low movement per bar is usually smaller but can fluctuate more frequently.
What changes across timeframes is not only the ATR number’s scale, but also the observation granularity and the smoothing effect of the lookback window.
Evidence or example (with explicit assumptions)
Assume you have access to the same underlying price process, and you compute ATR over the same lookback length in bars (for example, N bars), but with two different timeframes:
- Case A: ATR over N=14 bars on a shorter timeframe (each bar represents less time).
- Case B: ATR over N=14 bars on a longer timeframe (each bar represents more time).
Under these assumptions, both ATR values summarize “typical movement over N bars,” yet the meaning differs:
- In Case A, the 14 bars cover less total elapsed time, so ATR is more sensitive to recent changes in observed ranges.
- In Case B, the 14 bars cover more elapsed time, so ATR reflects a broader historical window and typically changes more slowly when volatility shifts.
This is the core time-sensitivity: ATR answers “how large were the recent bar-to-bar ranges in this timeframe,” so it will not match the volatility you experience if your observation timeframe and your holding period differ.
Limitations and risks
At least four material limitations can reduce ATR’s usefulness when timeframe and context are mismatched:
-
Past realized ranges, not future expectations ATR is derived from historical bars. A correct ATR computation can still fail to anticipate future volatility if volatility regime changes occur.
-
Regime shifts and burst behavior A short lookback or shorter timeframe may chase sudden spikes and then normalize quickly. A longer timeframe may dampen spikes and understate burst risk during the early part of a new volatility regime.
-
Bar construction and data handling ATR depends on how bars are built (for example, open/high/low/close boundaries) and what price feed is used. Comparing ATR values across providers can be misleading if the underlying bar data differs.
-
Costs and execution effects are not included ATR measures price range movement, not transaction costs, slippage, or execution quality. Two environments with the same ATR can produce different realized outcomes because of those non-price factors.
Verification and next question
You can independently verify timeframe sensitivity by recomputing ATR on the same data with different timeframes and/or different N values, and then comparing how quickly ATR changes after a known volatility increase or decrease.
A useful next question is: “Am I matching the ATR timeframe to the timeframe that best represents my decision or holding horizon?” If not, ATR may still be correct, but its interpretation can be disconnected from the volatility you actually observe.