Donchian Channel Breakout: The Turtle Rule Tested

Price moving beyond the boundary of the last 20 bars signals the start of a move worth following. This is the core of the famous turtle system, tested here in its pure form.

Donchian channel: the breakout rule

A close above the highest high of the previous 20 bars goes long; below the lowest low goes short. The position is held until the opposite signal, so the rule spends almost all its time in the market.

Results across 12 pairs

Run with the portal's own engine over 518 daily bars for each of 12 pairs, costs of 1 pip round turn, canonical parameters. The method and its limitations are set out in how we run backtests.

• Median return — -2.56%, profitable on 5 of 12 pairs
• Sharpe — -0.22, drawdown — 10.6%
• Trades — 9, win rate — 35%, time at risk — 95%
• Extremes: AUDJPY +24.2%, NZDUSD -18.0%

The widest spread of any rule tested

The median is slightly negative, but something else is more interesting: the spread between the best and worst pair is the widest of all ten rules we tested — over forty percentage points. On some pairs the rule delivered a substantial profit, on others an equivalent loss. That is trend-following behaviour in its pure form: it earns when a trend exists and bleeds methodically when one does not, and two years is too short for those regimes to average out.

What can and cannot be said about the turtle system

The rule can be called neither working nor broken on this sample — only that its result is determined entirely by whether a trend was present in the particular instrument. The original turtle system ran on a portfolio of dozens of markets over many years for exactly this reason: it relied on trends appearing somewhere. On one pair over two years that assumption does not hold. For comparison: simply holding the position on the same data returned 4.79%, and this rule did not beat it — why a benchmark is not optional.

What our test is missing

A caveat that matters specifically for a trend system: our run treats each pair separately, while the original approach was built on a portfolio of dozens of markets. The portfolio effect is the central mechanism of such systems, and it is absent from our test by construction. The general limitations — a short sample, correlated pairs, a signal tested without risk management — are listed in the methodology. The data is open and the run reproduces from a script in the repository.

This material is educational and is not individual investment advice. Backtested results do not guarantee similar results in the future. Trading forex carries the risk of losing capital.