Average True Range (ATR) is a volatility indicator: it estimates how much price has been moving over a chosen lookback. It does not tell you whether the next move should be up or down, which is why ATR is better treated as a risk and context tool than as a trade signal by itself (StockCharts ChartSchool: ATR, Fidelity: Average True Range, Charles Schwab: ATR and volatility).
That sounds useful because it is. It also sounds safer than it is. A volatility number can make a stop loss look scientific while the entry rule remains pure opinion.
What ATR actually is
ATR is a smoothed average of true range. J. Welles Wilder developed it as a volatility indicator, and both StockCharts and cTrader's own API documentation describe Wilder's recommended default as 14 periods (StockCharts ChartSchool: ATR, cTrader Algo: AverageTrueRange).
True range is designed to capture the current bar's range and any gap from the previous close. StockCharts and thinkorswim describe true range as the greatest of these three values: current high minus current low, current high minus previous close in absolute terms, and current low minus previous close in absolute terms (StockCharts ChartSchool: ATR, thinkorswim Learning Center: ATR).
true range = max(
high - low,
abs(high - previous close),
abs(low - previous close)
)
ATR = smoothed average of true rangeThe 14-period setting is a convention, not a law. thinkorswim lists 14-period Wilder smoothing as the default but also lets the user change both the length and moving-average type; cTrader exposes ATR as an indicator parameter rather than a fixed trading rule (thinkorswim Learning Center: ATR, cTrader Algo: AverageTrueRange).
How traders use ATR
Traders usually use ATR to normalize risk to current volatility. Schwab describes ATR as a way to put recent price movement into historical-volatility context, while OANDA and Fidelity both stress that ATR has no directional bias (Charles Schwab: ATR and volatility, OANDA: how to use ATR, Fidelity: Average True Range).
| Use | What ATR adds | What ATR cannot solve |
|---|---|---|
| Stop distance | A stop can be set relative to recent range instead of an arbitrary number of points | Whether the trade idea has positive expectancy |
| Position sizing | The same account risk can be translated into smaller size when the stop is wider | Whether the stop location is logically valid |
| Volatility filter | A system can avoid, reduce, or separate trades in unusually quiet or active regimes | Whether high volatility is bullish or bearish |
| Market comparison | ATR Percent can normalize ATR by price so different instruments are easier to compare | Whether two instruments have the same liquidity, spread, or execution quality |
The position-sizing use is the most practical. If a trader fixes account risk first, ATR can help convert volatility into trade size:
cash risk allowed / stop distance = position size
If stop distance is based on ATR, size falls when the market gets wider
and rises when the market gets quieter.The comparison point matters. Plain ATR is an absolute price amount, so it is hard to compare directly across differently priced instruments. Fidelity and StockCharts both describe Average True Range Percent (ATRP) as ATR divided by closing price and multiplied by 100, specifically to make volatility comparisons easier across securities (Fidelity: Average True Range Percent, StockCharts ChartSchool: ATRP).
What the critics get right
The cleanest criticism is that ATR is backward-looking. It summarizes recent range. It does not know why the range expanded, whether liquidity is thin, whether the next session has a news event, or whether spread will widen when the stop is hit.
The second criticism is that ATR can hide discretion behind a formula. "Two ATR stop" sounds objective, but the trader still chose the market, timeframe, ATR length, smoothing method, multiplier, entry rule, exit rule, session, and cost model. Change any one of those and the backtest can change.
The third criticism is that ATR is not a complete volatility model. Schwab frames ATR as historical volatility context, Fidelity says ATR is not directional, and OANDA explicitly warns that ATR should be paired with other analysis if the trader wants directional bias (Charles Schwab: ATR and volatility, Fidelity: Average True Range, OANDA: how to use ATR).
That does not make ATR useless. It makes it modest. ATR can help define risk in a way a computer can repeat. It cannot rescue a vague setup.
How you'd actually test it
To test ATR properly, do not test "ATR works." Test one ATR rule.
Start with a rule a machine can repeat:
- Define the market, timeframe, session, and data source.
- Define ATR length and smoothing method.
- Define the setup without hindsight language.
- Define the stop as a fixed ATR multiple or a named structure rule plus ATR buffer.
- Define position sizing from a fixed risk amount and the stop distance.
- Define exits, time stops, spreads, slippage, commission, and swap.
- Split in-sample and out-of-sample, then judge the rule on data it did not see during design.
Then compare variants:
| Test | Question |
|---|---|
| Same entry, fixed stop vs ATR stop | Does volatility-adjusted distance improve expectancy after costs? |
| Same entry, ATR sizing vs fixed lot | Does volatility sizing reduce drawdown without flattening return too much? |
| ATR filter on/off | Does avoiding very high or very low ATR regimes improve the trade distribution? |
| 14-period vs alternatives | Is the result stable, or did one parameter win by luck? |
This is the realbacktesting standard applied to an indicator: make the rule precise enough to fail. The discipline is the same as how to verify a cTrader backtest, fixed lot vs fixed risk for prop traders, and the broader proof mindset behind real-cost backtesting.
Frequently asked
Does ATR predict price direction?
No. ATR measures range or volatility, not direction. Fidelity, Schwab and OANDA all describe ATR as non-directional or focused on volatility rather than trend direction (Fidelity: Average True Range, Charles Schwab: volatility metrics, OANDA: how to use ATR).
Is 14-period ATR the correct setting?
14 periods is Wilder's common default, not a universal truth. StockCharts and cTrader both reference Wilder's 14-period recommendation, while thinkorswim makes the length and average type adjustable (StockCharts ChartSchool: ATR, cTrader Algo: AverageTrueRange, thinkorswim Learning Center: ATR).
Is an ATR stop automatically better than a fixed stop?
No. An ATR stop is more adaptive to recent range, but "adaptive" is not the same as profitable. The entry, exit, cost model, and out-of-sample result decide whether the rule has an edge.
Can ATR help position sizing?
Yes, if the sizing rule is defined before the trade. ATR can convert recent volatility into a stop distance, and the trader can size from allowed cash risk divided by that distance. The test still has to include real costs and drawdown behavior.
The stubborn takeaway
ATR is useful when it makes risk measurable. It is dangerous when it makes an untested opinion look measured.