Donchian Channel Tested: After a 20-Bar Breakout, EURUSD Went the Wrong Way First 57% of the Time, and 1 in 3 Breakouts Was Back Inside Within 3 Candles

Richard Donchian, often called the father of trend following, used the rule in the 1950s and 1960s: buy when price makes a new high for the last four weeks, sell when it makes a new low. The Turtle traders of the 1980s made the same rule famous. On a chart it is simply three lines: the highest high, the lowest low, and the middle between them.

The promise is also simple. A close above the top line means the market has escaped its recent range, so it should keep going. That is a claim about the next candles, so it can be counted.

What does the Donchian Channel measure?

The channel ignores closes, averages and volatility. It moves only when a new extreme is made, so it stays flat for long stretches and then jumps. On EURUSD M5 the 20-candle channel was a median 5.2 ATR wide, about 10 pips, and on gold M5 4.4 ATR.

One detail matters for every number below. The bands use the previous n candles, not the current one. If the current candle is included, a close can never be above the highest high, and a breakout by close becomes impossible. More on that in Donchian Channel settings for MT4.

How was it tested?

• Data: EURUSD M5 from June 3 to August 26, 2026, 16,926 bars; gold M5 from June 16 to August 26, 13,736 bars. M15 built from the same bars.
• Breakout: the first close above the upper band (or below the lower band) of Donchian 20. From that close, which came first within 48 bars, a 1 ATR move with the breakout or against it, compared with random bars and random sides. A breakout is “false” if any of the next 3 closes is back inside.
• Five rules, Donchian 20: the breakout; the breakout on the EMA 200 side only; a fade that takes the opposite side of the breakout; a close crossing the middle line; and a squeeze breakout, taken only when the channel was in the narrowest quarter of the last 200 bars.
• Exit and pass: stop 1.5 ATR, target 3 ATR, 48 bars at most, one trade at a time, the same exit as in the forex indicators list, ranked. Cost 1 pip on EURUSD and $0.30 an ounce on gold, from spread and swap costs in MT4. A run passes if it makes money after cost, in both halves (split July 15), and beats the 95th percentile of 200 random-entry runs with the same exit, as in algorithmic rule vs random entry.

How often does price break the Donchian Channel?

One bar in ten closed outside a 20-candle channel, on both pairs and both timeframes. On a 5-minute chart that is about 21 fresh breakouts a trading day. A breakout is not a rare event, and rare events are the only kind that can carry a strong message.

Many of them did not last. On EURUSD M5 a third were back inside within 15 minutes, and on gold almost half. The same fast snap-back showed up for stop hunts in liquidity sweeps tested.

Did price keep going after a breakout?

On EURUSD the breakout pointed the wrong way. On M5, 57 breakouts in 100 hit 1 ATR against you before 1 ATR in your favour, and 12 candles later price was on average 0.3 ATR back on the other side. That is 10 points worse than a random bar with a random side.

Gold was different. There the breakout was about a coin flip on M5 and a little better than random on M15, 50.0% against 45.2%, and price was still 0.4 ATR further on 12 candles later. Gold trended, EURUSD went back and forth, the same split as in gold vs EURUSD with one rule.

Did any Donchian rule make money?

Pips on EURUSD, dollars an ounce on gold, trades in brackets. The random rows match the breakout’s trade count; every run was checked against its own random range.

No run passed all three checks. Four of twenty made money after cost, all on gold M15. The breakout there earned +$2.07 a trade and was in profit in both halves, +$2.65 and +$1.81. But random entries with the same stop and target reached +$2.99 at the 95th percentile on that chart, because gold rose so much in those weeks that almost any long did well. The breakout did not beat that.

On EURUSD M5 every rule lost, and the breakout lost more than the median random entry, -1.45 against -1.00 pips. Adding the EMA 200 filter barely changed it. Filters rarely rescue a signal that points the wrong way, as in chart patterns vs range breakout.

Why was the squeeze breakout the worst?

The common advice is to wait for a narrow channel and trade the breakout from it, because quiet markets are supposed to explode. On EURUSD M5 that was the worst rule in the test: -2.18 pips a trade, 23.6% winners, and in profit in neither half. After a squeeze breakout price hit 1 ATR against the breakout first 61.6% of the time.

The reason is in the width. A narrow channel means a small range, so the breakout level is close and easy to touch. A tiny push above a quiet range was not the start of a move on EURUSD, it was noise at the edge of the range. On gold M15 the squeeze version made +$1.12 a trade, but on 79 trades, and it lost in the first half.

Is fading the breakout better?

On three of the four charts the fade lost less than the breakout; only on gold M15 did following it do better. On EURUSD M5 it was in profit before cost, +0.41 pips a trade, and lost 0.59 after the spread, which is better than 95% of random-entry runs, -0.70. So on EURUSD the honest reading of a 20-candle breakout is “probably a short-term top”, not “a trend has started”.

It still did not pay. A gross edge of 0.4 pips does not cover a 1-pip spread, and the fade lost in both halves. The best settings in the grid of 24 are in the MT4 settings guide.

What would I do with the Donchian Channel?

• Do not trade a 20-candle breakout on EURUSD M5 or M15. It went against you first more often than a random entry.
• Treat the squeeze breakout with extra caution. On EURUSD it was the worst of the five rules, not the best.
• Use the channel to see the range, not to time an entry. The top and bottom lines show where the recent extremes are. That is useful for placing a stop, as in fixed vs ATR stop.
• If you want the Turtle version, test the exit too. The Turtles did not use a fixed target. How their channel exit changed the result is in Donchian breakout vs Turtle rules.

Frequently asked questions

What is a Donchian Channel?

Three lines on the chart: the highest high of the last n candles, the lowest low of the last n candles, and the middle between them. The usual period is 20.

Does the Donchian breakout work on forex?

Not on EURUSD in this test. After a 20-candle breakout, price moved 1 ATR against it first 55% to 57% of the time on M5 and M15, and the breakout lost 0.57 to 1.45 pips a trade after the spread.

How many Donchian breakouts are false?

On EURUSD 33% to 38% of breakouts closed back inside the channel within 3 candles. On gold it was 48% to 49%.

Is the Donchian squeeze breakout a good strategy?

Not on EURUSD M5, where it was the worst rule tested: -2.18 pips a trade with 23.6% winners. On gold M15 it made money but on only 79 trades, and it lost in the first half.

Which market did the Donchian breakout work on?

Only 15-minute gold made money, +$2.07 an ounce a trade, but random entries with the same exit did as well or better on that chart.