ATR Volatility Breakout Tested: Following the Break Lost Before Costs in 289 of 310 Setups, and the Bigger the Break, the Worse

The volatility breakout is one of the oldest ideas in trading. Measure how far the market normally moves with the Average True Range, and when price moves much further than normal, jump in: the move is “real”. It is the logic behind Keltner Channel breakouts, range-expansion bars and Larry Williams’ daily breakout. On this blog ATR has so far only been a stop, in my fixed vs ATR stop test. Here it is the entry.

The short answer: on EURUSD 5-minute bars an unusually large move was more likely to be given back than extended. Following it lost almost every time. Fading it had a small edge that the spread mostly ate.

What is an ATR volatility breakout?

ATR is the average size of a bar, counting gaps: the true range of each bar, averaged over N bars with Wilder’s smoothing. On EURUSD M5 the default ATR 14 had a median of 1.96 pips. It sat between 1.26 and 3.22 pips on 80% of bars. A breakout rule says: when price moves k times that amount, trade the direction of the move.

How was the breakout tested?

• Data: 16,926 EURUSD M5 bars, June 3 to August 26, 2026, split at July 15 into two halves.
• Impulse: the close is more than k×ATR above or below the previous close.
• Range expansion: the bar’s true range is more than k×ATR; trade in the bar’s colour.
• Keltner channel: the close crosses outside a 20-period EMA ± k×ATR.
• Daily (Larry Williams style): the first close more than k × yesterday’s range away from today’s open. k = 0.25, 0.5, 0.75, 1.
• Grid: ATR 7, 14, 20 and 50; k = 1, 1.5, 2, 2.5 and 3.
• Exits: ATR stop and target at 1:1, 1:2, 1.5:3 and 2:2 (capped at 4 hours), or a plain exit after 2 hours.
• Cost: 1 pip per trade, as in my spread and swap test. One trade at a time. Where stop and target sat inside the same bar, I counted the stop.
• Mirror: every setup was also traded the other way — against the break.

That is 640 setups, 620 of them with at least 30 trades. Only those 620 are counted below.

Does following the breakout work?

No. Here is every family, follow and fade side by side:

Pips per trade before the 1-pip cost, EURUSD M5.

After the 1-pip cost the picture is flat on both sides: 1 of 310 follow setups made money, and 13 of 310 fade setups. The only follow setup in profit was the daily breakout at k = 0.25 with a 2-hour exit, +1.57 pips over 46 trades — +3.27 in the first half and +0.14 in the second.

Did bigger breakouts work better?

The opposite. The textbook says a bigger move is a stronger signal. Here, the bigger the multiplier, the more following it lost:

Median over 20 setups per row: 4 ATR periods × 5 exits.

The raw price action says the same thing without any exit rule. I sorted every bar by its true range in ATRs and looked at the next 12 bars — one hour:

Pips in the direction of the bar’s colour. The rest of the time the close an hour later was against the bar or unchanged.

A bar three times the normal size was followed by a move in the same direction only about one hour in three, and on average price gave back 1.28 pips. On a 5-minute EURUSD chart a big bar was usually the end of a push, not the start of one. It is the same thing I saw in the Bollinger Bands test on M5, where fading the band beat following it.

Which exit was best?

None of them rescued the follow side. Median net pips per trade:

The ATR period made almost no difference: ATR 7, 14, 20 and 50 had median net results between -1.22 and -1.30 on the follow side and -0.59 to -0.63 on the fade side. The multiplier and the direction mattered. The period did not — the same thing my optimized vs default settings test found for other indicators.

Why does the fade still lose after costs?

Because a 1-pip spread is half an ATR on this chart. With ATR 14 at a median of 1.96 pips, a 1:1 ATR target is a 2-pip target, and the spread takes half of it before the trade starts. The Keltner fade at 2×ATR with a 1:2 exit shows the problem clearly: 919 trades, +0.39 pips each before costs, -0.61 after, and -0.53 and -0.69 in the two halves. Consistent, and consistently smaller than the cost.

The setups that did clear the spread used bigger moves or longer holds:

The daily fade is the only result large enough to matter, and it rests on 36 trades over 12 weeks — one on most days. I would not trade it on this evidence. I would forward test it, because at this multiplier the fade was positive before costs with all five exits, and the follow side of the same rule lost 5.89 pips per trade.

Does the session change the answer?

A little. For the Keltner 20 / 2×ATR breakout, following it lost the least in London hours (07:00–11:59 UTC): +0.45 pips before costs and -0.55 after, on 212 trades. Fading it did best late in the day (17:00–23:59 UTC): 60.8% winners, +0.37 before costs. Neither cleared the spread. My session vs all-day test covers why time filters rarely save a rule that loses all day.

So how would I use an ATR breakout?

• Do not buy a big 5-minute bar on EURUSD because it is big. Following it lost in 309 of 310 setups after costs.
• Use ATR to size, not to enter. A 1.96-pip ATR tells you that a 2-pip target is mostly spread. For stops, my stop distance test is the better guide.
• If you fade, fade the big ones. The fade edge grew from +0.27 to +0.69 pips as the move grew from 1× to 3×ATR.
• Move up a timeframe. On M30 and H1 the ATR 14 is 6 to 9 pips, the spread shrinks to a sixth or a ninth of it, and the same Keltner fade cleared the spread in both halves in several settings. Higher charts also helped the trend rules in my Supertrend settings test.

Frequently asked questions

Does the ATR breakout strategy work on forex?

Not on EURUSD 5-minute bars. Trading in the direction of a move larger than k×ATR lost before costs in 289 of 310 setups and after a 1-pip cost in 309 of 310.

What ATR multiplier is best for a breakout?

For following the move, none: every multiplier from 1 to 3 lost, and larger ones lost more, from -0.20 pips per trade at 1×ATR to -0.58 at 3×ATR before costs. For fading it, larger was better.

Do big candles continue on EURUSD M5?

Usually not. After a bar at least three times the ATR, the next hour closed in the same direction 36.2% of the time, and price gave back an average 1.28 pips.

What is a normal ATR for EURUSD on the 5-minute chart?

ATR 14 had a median of 1.96 pips from June to August 2026, and stayed between 1.26 and 3.22 pips on 80% of bars.

Does the Larry Williams volatility breakout work on EURUSD?

Following it did not: at half of yesterday’s range it lost 5.89 pips per trade with a 2-hour exit. Fading the same breakout made +3.89 pips per trade after costs, but on only 36 trades.