Donchian Breakout vs Turtle Trading Rules: The Turtle Exit Lost Less on Both 5-Minute Charts, More on Both 15-Minute Charts, and 10% of Trades Made 64% to 89% of the Profit

In 1983 Richard Dennis and William Eckhardt trained a group of beginners, the Turtles, to trade a fixed set of rules. The entry was a Donchian Channel breakout. The exit was the part people forget: no profit target at all, just a trailing channel on the other side and a 2 ATR stop. Most trades lost small, and a few trends paid for everything.

On this blog every indicator test uses the same fixed exit, a 1.5 ATR stop and a 3 ATR target. The full Donchian test with that exit is in Donchian Channel tested: the breakout went against you first more often than a random entry on EURUSD. The fair question is whether the Turtle exit changes that.

What were the Turtle rules?

The Turtles traded daily charts of futures, with position sizing and a filter that skipped a System 1 trade after a winner. I kept only the entry, the exit and the stop, one trade at a time, on intraday charts. That is the version most forex traders actually try in MT4. The cost was 1 pip on EURUSD and $0.30 an ounce on gold, as in spread and swap costs in MT4.

How was it compared?

• Data: EURUSD M5, 16,926 bars from June 3 to August 26, 2026; gold M5, 13,736 bars from June 16; M15 built from the same bars.
• Same entries: every breakout of System 1 and System 2 was traded twice, once with the Turtle exit and once with the fixed exit (stop 1.5 ATR, target 3 ATR, 48 bars).
• Noise check: the difference per trade was resampled 2,000 times. If the 90% range crossed zero, the difference is called noise.
• Halves: results before and after July 15, so one lucky stretch cannot carry the whole number.

Did the Turtle exit beat a fixed target?

Pips on EURUSD, dollars an ounce on gold, trades in brackets. The Turtle exit traded less because it stayed in the market longer.

The pattern is clean and a little strange. On both 5-minute charts the Turtle exit did better with both systems. On gold M5 it turned a loss into a profit: System 1 from -$0.90 to +$0.24, System 2 from -$1.20 to +$0.67. On both 15-minute charts it did worse with both systems.

The resampling did not back any of the eight differences. The largest, System 2 on gold M5, was +$1.87 a trade, but its 90% range ran from -$0.65 to +$4.52. On 215 trades with a few big winners, that is too wide to call.

What did the Turtle exit do to the win rate?

This is what trend following feels like. With the Turtle exit, three trades out of four lost on EURUSD M5, and the money came from a handful of long runs: the best tenth of trades produced 82% of all the profit there. System 2 was even more lopsided, 88% to 89% on EURUSD and 21% to 22% winners.

That has a practical cost. A rule that wins one trade in four needs long losing streaks to be survivable, which is why the Turtles sized positions so carefully. The same trade-off, small certain wins against rare large ones, is in trailing stop vs take profit.

Why did the Turtle exit help on M5 and not on M15?

A fixed 3 ATR target caps every winner at twice the stop. On the 5-minute charts the breakout itself was close to useless, so the only way to make money was to catch the occasional long run, and the Turtle exit let those runs go. On EURUSD M5 it lifted the result before cost from -0.45 pips to -0.13 with System 1 and to -0.02 with System 2.

On M15 an ATR is about twice as many pips, and the fixed target already reached far. The 10-bar and 20-bar exits then gave back more of each move before they closed it. On gold M15 the plain breakout with a fixed exit was the best of the five default Donchian rules, +$2.07, and the Turtle exit only lowered it to +$1.77.

On EURUSD the exit could not fix the entry. Even the best Turtle version, System 2 on M5, broke even before cost and lost a pip a trade after it. A better exit only helps an entry that has something to give, as in recognizing exit signals with an indicator.

Was one Turtle system better?

System 2 traded 40% to 47% less than System 1 and won more on gold M5, +$0.67 against +$0.24, in profit in both halves, +$0.43 and +$0.80. On the other three charts it was worse or close. Slower settings cut the number of false breakouts but also cut the signals, the same lag trade-off as in indicator lag vs whipsaw. Neither system made money on EURUSD on either timeframe.

What would I take from the Turtle rules?

• Test the exit, not just the entry. The same breakout went from -$1.20 to +$0.67 a trade on gold M5 by changing only the exit.
• Do not expect the Turtle rules to work on EURUSD intraday. All four EURUSD runs lost 1.0 to 2.3 pips a trade after the spread.
• Expect to lose most trades. 21% to 40% winners, with most of the profit from a tenth of the trades.
• Do not choose the period from one sample. How 24 Donchian settings ranked from one half to the next is in Donchian Channel settings for MT4.

Frequently asked questions

What are the Turtle trading rules?

System 1 buys a close above the 20-bar high and exits on a close below the 10-bar low; System 2 uses 55 and 20 bars. Both use a 2 ATR stop and no profit target. Shorts are the mirror.

Do the Turtle trading rules still work?

On intraday EURUSD, no: all four runs lost after a 1-pip spread. On gold 3 of 4 runs made money, but none of the improvements over a fixed exit was outside the noise.

Is a Donchian exit better than a fixed take profit?

It was better on both 5-minute charts and worse on both 15-minute charts, for both systems. 4 of 8 comparisons went its way.

What is the win rate of the Turtle system?

21% to 40% in this test, against 30% to 41% for the same entries with a fixed exit. The best 10% of trades made 64% to 89% of the profit.