ATR Explained: Set Your Stop Loss and Lot Size by Market Volatility

ATR (Average True Range) is the average of how wide price swings per bar. J. Welles Wilder introduced it in his 1978 book. ATR doesn’t point in any direction. It tells you “how wide is a normal swing in this market right now”, which is exactly what you need every time you set a Stop loss or work out a lot size. This article covers the formula, a worked example, and four ways to use ATR to manage risk.

True Range: bar width that counts gaps

If you measure a bar only by High minus Low, you miss the times price jumps (a gap), such as a Monday open far from Friday’s close, or a bar right after news. Wilder used True Range (TR) instead, which takes the largest of three values:

1. This bar’s High minus its Low
2. This bar’s High minus the previous close (ignoring the sign)
3. This bar’s Low minus the previous close (ignoring the sign)

Example with three made-up bars:

Bar	Previous close	High	Low	H − L	|H − prev close|	|L − prev close|	TR
A. normal	2,400	2,406	2,398	8	6	2	8
B. gap up	2,404	2,421	2,415	6	17	11	17
C. gap down	2,418	2,410	2,401	9	8	17	17

Bar B’s body is only 6 wide, but price really travelled from 2,404 up to 2,421. Anyone holding a Sell took a 17-point move, not 6, so TR counts 17. Bar C is the same thing the other way.

From TR to ATR: Wilder’s smoothing

ATR is the average of past TR values. The default is 14 bars. Wilder’s averaging works like this:

• First value = the simple average of the first 14 TRs
• Each next value = (previous ATR × 13 + latest TR) ÷ 14

Example: the previous ATR is 9.00, then bar B arrives with a TR of 17.

New ATR = (9.00 × 13 + 17) ÷ 14 = (117 + 17) ÷ 14 = 134 ÷ 14 = 9.57

A single bar almost twice as wide lifts ATR by only about 6%, and its effect then fades a little at a time. With a simple average, that bar would stay in the formula for exactly 14 bars and then drop out all at once. This is the same smoothing as the SMMA (or RMA) line, which RSI also uses.

ATR is measured in the asset’s own price units. A gold ATR of 6 means an average swing of 6 dollars per bar. A currency pair ATR of 0.0012 is 12 pips. It always depends on the timeframe: ATR on an M5 chart and on an H4 chart differ several times over.

The ATR that ships with MT4/MT5 takes a simple average (SMA) of TR rather than Wilder’s smoothing, while TradingView uses Wilder’s by default. The two won’t match exactly, especially after a big bar. Either works. Just stick with one.

ATR measures volatility, not direction

ATR tells you how wide price is swinging, not where it is going. A rising ATR is not a Buy signal and not a Sell signal.

A long stretch of low ATR says the market is quiet, which is sometimes followed by a strong move, but it doesn’t tell you when or in which direction. If volatility squeezes interest you, read Bollinger Bands. To measure how strong a trend is, use ADX, which Wilder built with ATR inside its formula.

Setting a Stop loss with ATR

The idea is that a Stop loss should sit outside the range price swings through normally. Place it tighter than the normal swing and you get knocked out often, even when your direction was right. Common distances are 1.5 to 3 times ATR on the timeframe you trade:

• 1.5× tight, suits entries with clear structure close by
• 2× a middle value many traders start with
• 3× wide, survives a lot of noise, often used for holding with a trend

These numbers are rough guides, not values proven to be best. A good approach is to combine them with chart structure: place the stop behind the previous low, then check that the distance isn’t tighter than 1.5 ATR. If it is, a normal swing alone may reach it. More on this in Wide or tight Stop loss.

The benefit of an ATR stop is that it adapts on its own: tighter in a quiet market, wider in a volatile one. A fixed stop (say 20 pips on every trade) is too tight on volatile days and too wide on quiet ones.

ATR trailing stop: follow price, keep a volatility-sized gap

A trailing stop is a Stop loss that moves along as price goes your way. The best-known ATR version is Chuck LeBeau’s Chandelier Exit:

• Holding a Buy: stop = highest High of the last 22 bars − 3 × ATR(22)
• Holding a Sell: stop = lowest Low of the last 22 bars + 3 × ATR(22)

Example: you hold a Buy, the highest High of the last 22 bars is 2,450 and ATR(22) = 8. Stop = 2,450 − 3 × 8 = 2,426.

The key rule is that the stop only moves in the direction that locks in profit. If the calculation comes out lower than before (for example, ATR swells in a volatile spell), keep the old value. Never move it back. A stop like this never exits at the top. It always gives back part of the gain before it is hit, in exchange for staying in long trends.

Sizing the lot from an ATR stop

An ATR stop gets wider on volatile days. Keep the same lot and the money lost per trade grows with it, so recalculate the lot every trade, using the formula in Position size from Stop loss.

Example: a 5,000-dollar account risking 1% = 50 dollars per trade, trading gold (1 lot is usually 100 ounces, so a 1-dollar move = 100 dollars per lot). The ATR values are made up.

	Normal day	Volatile day
ATR(14)	6 dollars	10 dollars
Stop = 2 × ATR	12 dollars	20 dollars
Loss per 1 lot if stopped	12 × 100 = 1,200	20 × 100 = 2,000
Lot = 50 ÷ loss per 1 lot	0.0417 → 0.04	0.025 → 0.02
Actual loss if stopped	4 oz × 12 = 48	2 oz × 20 = 40

Always round the lot down, so the actual loss stays under 50 dollars on both days. The stop got wider but the risk didn’t grow, because the lot got smaller. Try your own numbers with the lot calculator at the end of this article.

Volatility filter: ATR too small compared with spread

Spread is a fixed toll per trade. It doesn’t shrink with ATR. When ATR is very small, spread takes a big share of the distance price can travel. An example from our data: ATR on the gold M1 chart in 2023 was about 0.48 dollars, while the gold spread we measured was about 0.27 dollars. So spread equalled roughly 56% of an average bar’s swing. A stop of 1.4× ATR would be only about 0.67 dollars away, and spread eats about 40% of that the moment you open. Compare a made-up H1 ATR of 6 dollars, where the same spread is only 4.5%. The full picture is in Spread cost on small timeframes.

To use this, set a rule in advance, for example: don’t trade if spread is more than 10% of ATR on your timeframe. The 10% is an example, not a standard. Test it on your own system.

In our backtests, an ATR filter helped in some cases, but not always:

• A system that trades breakouts from a 5-bar box on gold M1, 2026, real tick data, 0.01 lot. With no filter it averaged −0.189 dollars per trade. Requiring the box to be wider than 3 × ATR(100) brought that to −0.042 dollars per trade. Much smaller losses, but still losses. A volatility filter alone didn’t create an edge.
• A system that enters on pullbacks to the Fibonacci 61.8% level on gold M1, 0.10 lot, 10,000-dollar account, trading only when ATR on the H1 chart was above its 20-day average. In 2026 (real ticks), profit stayed close to the same (+5,373 versus +5,320 dollars unfiltered), but the Profit factor rose from 1.44 to 2.18 on just 86 remaining trades. It looked like the filter cut the bad trades well. But in 2025 profit fell from +2,091 to +1,260, and in 2024 it went from +1,178 to −101 dollars (both years use 1-minute price data). The filter only helped in 2026, so we don’t use it.

These are backtest results. Past results do not predict future results, and they are not returns you should expect. A filter that looks good in a single year is a common trap. Read more in Overfitting: choose the plateau.

Comparing volatility across markets with ATR as a percentage

ATR is in price units, so you can’t compare it across markets directly. A gold ATR of 40 dollars doesn’t mean gold is six thousand times more volatile than a currency pair with an ATR of 0.0065. Divide by price first:

ATR% = ATR ÷ price × 100

Example with made-up numbers on the daily chart:

Market	ATR(14)	Price	ATR%
EURUSD	0.0065	1.0800	0.60%
Gold	40	2,600	1.54%
Bitcoin	2,400	60,000	4.00%

With these numbers, Bitcoin swings about 6.7 times as much as EURUSD per day in proportional terms. Using a 2 ATR stop on both gives stop distances in % that differ by the same ratio, and the lot sizes have to differ a lot to carry the same risk. ATR% also helps compare one market over time, because a higher price makes the dollar ATR grow on its own even if real volatility is unchanged.

When ATR misleads

• After big news or a gap, ATR swells and stays high for several bars. Stops worked out in that period will be wider than the swings that have already calmed down.
• Before news, markets are often quiet with a low ATR, so an ATR stop is tight just when price is about to jump. Always check the economic calendar first.
• Using ATR from the wrong timeframe: entering on M5 but using H4 ATR, or the other way round, puts the distance off by several times.
• Tuning the period on a single year: in our backtest, switching a box filter on gold M1 from ATR(100) to ATR(14) made 2026 alone about 11% more profitable, but over 5 years (2022–2026, older years using ticks simulated from 1-minute data) the total came out about 5% lower. Backtest results do not predict future results.

Common mistakes

• Reading a high ATR as a trade signal. ATR has no direction. A sharp drop also pushes ATR up.
• Using an ATR stop without cutting the lot. On a volatile day the stop doubles and so does the money at risk. Recalculate the lot every trade.
• Using ATR alone without structure. A 2 ATR stop that lands right in the middle of a range can be hit more easily than a slightly wider one behind the previous low.
• Forgetting spread on small timeframes. ATR on M1 can be so small that spread takes half of a normal swing.
• Comparing raw ATR across markets. Convert to a percentage of price first.
• Moving a trailing stop back when ATR swells. Once a stop has locked in profit, it must not move back.

ATR pairs well with indicators that show momentum or direction, because they measure different things. See how to combine them without overlap in Combining indicators without redundancy, and if you trade breakouts, ATR fits naturally with the Donchian Channel.