ATR & Volatility

Average True Range (ATR) measures how much an instrument moves, not which way it moves. Wilder introduced it as a volatility statistic: for each bar, the *true range* is the largest of three distances — the bar's own high minus low, the distance from the previous close to the current high, and the distance from the previous close to the current low. The last two terms exist so that gaps and overnight moves count as range instead of being ignored. ATR is then a moving average of that value, classically Wilder-smoothed over 14 periods. The output is expressed in the instrument's own price units, which is why ATR is used as a unit of distance rather than as a signal.

That distinction defines the concept. ATR has no directional bias and no overbought or oversold reading; on its own it produces no trades. Its job is to establish what counts as a large move right now, so that stops, targets, position size, filters and channel widths can be expressed in multiples of current volatility instead of fixed points or percentages. A 20-point stop is arbitrary; a 1.5x ATR stop holds the same relative distance whether the market is quiet or fast, and adapts as conditions change without re-optimisation.

## How ATR enters a strategy

Four roles cover most implementations, and many systems use more than one at once.

**Risk distance.** Stops and targets placed at a multiple of ATR from the entry price, or from a swing point rather than from the fill. This is the most common use, and the reason ATR appears inside strategies whose entry logic has nothing to do with volatility.

**Position sizing.** Size derived from a fixed risk amount divided by the ATR-based stop distance, so risk per trade stays roughly constant in currency terms while the stop breathes with the market. This has the largest effect on the shape of an equity curve, and it is the part most often left implicit in a video explanation.

**Trailing and exit management.** Chandelier-type trails, ATR bands and swing-style trend lines that step behind price at a volatility-scaled offset. The multiplier decides how much noise a trade tolerates before it is closed, so it acts as a trend-following aggressiveness dial rather than a cosmetic setting.

**Filters and channels.** Keltner Channels (a moving average plus and minus an ATR multiple), volatility-expansion breakouts, range-contraction patterns such as NR7, and regime filters that stand aside when ATR is unusually low or unusually high relative to its own recent history. Here ATR becomes part of the entry condition.

## What differentiates implementations

Versions of the same nominal idea usually diverge on a short list of choices. **Smoothing and period:** Wilder's average over 14 bars is the reference, but SMA and EMA variants and periods from 5 to 50 behave measurably differently, with shorter lookbacks reacting faster and cycling stops more often. **Multiplier:** roughly 1x to 4x spans scalping through position trading, and it interacts with the timeframe rather than being independent of it. **Anchor:** ATR measured from the entry, from the signal bar, from the highest close since entry, or from a structural level. **Normalisation:** raw ATR is in price units and is not comparable across instruments, so some systems use ATR as a percentage of price or as a score against its own distribution. **Bar reference:** whether the ATR value is read from the last closed bar or from the still-forming one — a frequent source of look-ahead in indicator code.

## Common mistakes

Treating ATR as a direction signal is the first: rising ATR says the market is moving, not that it is moving up. Comparing raw ATR values across symbols or across long spans without normalising is the second — a reading of 50 means nothing without the price level it belongs to. Others recur often: tuning the multiplier on the same data used to judge the outcome; setting stops so tight in ATR terms that they sit inside the spread or typical slippage; pairing an ATR stop with an ATR trail whose parameters contradict each other; and, intraday, ignoring that ATR has a seasonal shape within the session, so one multiplier applies a much tighter or looser stop at the open than at midday.

## Evaluating a version

Because ATR usually governs exits and sizing rather than entries, test it as a risk layer. Run the entry logic with a fixed stop as a baseline, then substitute the ATR version on identical entries: if the volatility scaling does not change the result, it is not doing the work the description claims. Examine the multiplier as a surface rather than a point — a setting that works only at 2.1 and degrades at 1.9 and 2.3 is fitted. Include realistic costs, since ATR-based exits change trade duration and turnover. Segment results by volatility regime (calm, trending, shock) instead of reading aggregate statistics alone, and confirm the calculation uses closed bars in whichever platform you use, since identical rules give different results in a repainting implementation. Walk-forward or out-of-sample testing matters more here than in most concepts, because volatility is itself regime-dependent and a multiplier fitted to one regime rarely carries into the next unchanged.

Strategies in this concept (102)

Frequently asked questions

Does ATR tell you when to buy or sell?

No. ATR is a non-directional volatility measure — it reports the size of recent price movement, not its direction. A rising ATR means the market is moving more, which can happen in an advance, a decline or a two-sided shakeout. Every strategy in this catalogue that uses ATR pairs it with a separate directional component (trend, structure, momentum, pattern or channel logic); ATR then sizes the risk, the stop or the filter around that decision.

What ATR period and multiplier should I use?

There is no universal pair, and any specific numbers should be treated as a starting point tied to a timeframe and instrument. Period 14 with Wilder smoothing is the reference implementation; shorter periods react faster and produce more stop movement, longer periods are steadier but slower to adjust after a regime change. Multipliers roughly between 1x and 4x cover scalping through swing use. The practical test is stability: check whether performance holds across a neighbourhood of values rather than at one exact setting.

Can I compare ATR values between different markets?

Not in raw form. ATR is denominated in the instrument's price units, so a reading on an index future, a currency pair and a crypto pair are not on the same scale — and they are not even comparable for the same symbol across years if price levels have changed substantially. To compare, normalise: express ATR as a percentage of price, or as a ratio against its own longer-term average. Multiplier-based rules avoid the problem implicitly, since they are relative by construction.

Is an ATR stop better than a fixed stop?

It is a different trade-off, not automatically better. An ATR stop keeps the exit distance proportional to current conditions, which usually reduces premature exits in fast markets and tightens risk in quiet ones. The cost is that stop distance — and therefore per-trade risk in currency terms — varies unless position size is adjusted alongside it. The comparison worth running is the same entry logic with both stop types, with costs included, so the volatility scaling can be judged on its own contribution.

How does ATR relate to Bollinger Bands and other volatility measures?

Both describe dispersion, but differently. Standard deviation (used by Bollinger Bands) is computed from closing prices and reacts to how spread out closes are; ATR is computed from bar ranges and includes gaps, so it captures overnight moves and limit-style openings that a close-to-close measure understates. Keltner Channels are the ATR analogue of Bollinger Bands. In practice ATR is preferred where the concern is the distance price can travel against a position, and standard deviation where the concern is statistical deviation from a mean.

What should I check before trusting a backtest of an ATR-based version?

Confirm the ATR is calculated on closed bars, since reading the forming bar introduces look-ahead that inflates results. Verify that stop distances remain larger than realistic spread and slippage on the instrument and timeframe. Check that the multiplier and period were not tuned on the same data used to report performance, and inspect results separately across calm, trending and high-volatility periods — an ATR layer that only helps in one regime will not generalise, and aggregate statistics hide that.

All strategy concepts · Explore all strategies