Donchian Channel indicator: settings and signals
The Donchian Channel is simply the high and low over N bars, i.e. the levels whose breakout means a new extreme of the period.
What the indicator looks like on a real chart
Data: EUR/USD, daily bars, 2025-06-16 — 2026-07-20. The indicator is computed over 400 bars; the chart shows the last 180. Quote source: Binance Spot REST API (api.binance.com/api/v3/klines).
| Regime | Score |
|---|---|
| Trending market | 63 |
| Range | 28 |
| High volatility | 52 |
| Low volatility | 35 |
What the indicator actually measures
Of all the indicators in the set this is the most direct: no averaging or smoothing, just two extremes. Hence the characteristic stepped shape of the bounds — they stand still until an extreme is updated and change in a jump when it happens.
It was on this construction that Richard Dennis built the Turtle system — one of the few publicly documented profitable mechanical approaches. The logic is simple: if price has gone beyond the high of the last N bars, something new is happening and it is worth following the move.
The flat sections of the bounds matter more than the breakouts themselves. A long horizontal shelf means the market has not updated an extreme for a long time, i.e. it is in a squeeze. A breakout of such a shelf is more meaningful than a breakout of a bound that was already moving every bar.
Formula
The upper bound is the highest high over N bars, the lower is the lowest low, and the middle line is their half-sum. No weights, smoothing or derived quantities: the bounds are defined by two specific bars in the window.
Standard settings
| Parameter | Value |
|---|---|
| Classic breakout period | 20 |
| Exit on the opposite channel | 10 |
| Higher period (Turtle System 2) | 55 |
Period 20 for entry and 10 for exit is the configuration from the Turtle system: entry on the longer window, exit on the shorter one, to lock in a reversal faster. A period of 55 was used there too as a slower variant for large moves.
Implementation code (Python)
def donchian(high, low, period=20):upper = high.rolling(period).max()lower = low.rolling(period).min()return upper, (upper + lower) / 2, lower
When the indicator stops working
- The bounds are defined by a single extreme bar, so one spike sets the level for the whole window period.
- In a range, breakouts of the bounds systematically turn out to be false.
- The indicator says nothing about the strength of a breakout — a breach of the bound by one pip and by a hundred look the same.
Common mistakes in use
- Trading a breakout without a market-state filter. Breakout strategies are unprofitable in a range, and this is their main property, not a defect of the implementation.
- Using the same period for entry and exit — then the system reverses too late and gives back a significant part of the profit.
- Forgetting that the level is set by a single bar: an anomalous news spike will form a bound that the market may never approach again.
- Expecting a high share of winning trades. Channel-breakout systems usually win in a minority of trades and make their profit from rare large moves.
Who needs this and why
For trend-following traders prepared for a low share of winning trades in exchange for a few large moves. It requires discipline: most signals will be losing, and that is built into the approach.
Frequently asked questions
Why do the channel bounds stay put for several days?
Because a bound is a specific extreme over the period, and it changes only when a new high or low appears or when an old one leaves the window. A flat bound means the market has not updated an extreme for a long time, i.e. it is in a squeeze.
Which period should I choose for entry and exit?
The classic Turtle configuration is 20 bars for entry and 10 for exit, plus the slower pair of 55 and 20. The point is the asymmetry: you need to exit faster than you enter, otherwise the position is held too long against a reversal that has begun.
Why do breakout strategies have so many losing trades?
It is a structural property of the approach. Most breakouts never grow into a trend, so the share of winning trades is usually below half. A positive result is achieved through size asymmetry: losses are capped by the stop, while the rare successful entries are held in a prolonged move.