Donchian Channel Explained: Trading Breakouts the Turtle Way

The Donchian Channel is a price channel whose upper line is the highest High and whose lower line is the lowest Low of the last N bars. There is no averaging or weighting. When price leaves the channel, it has made a new high or low for the last N bars. That is the core of breakout trading and of the Turtle rules, one of the best-known stories in trend following. This article covers the formula, how to read it, the classic rules, ATR-based stops and why most breakouts fail.

The Donchian Channel formula

Line	Formula
Upper	Highest High of N bars
Lower	Lowest Low of N bars
Middle	(Upper + Lower) ÷ 2
Channel width	Upper − Lower

It is named after Richard Donchian, a pioneer of trend following. The most common setting is N = 20, the default for Donchian Channels (DC) on TradingView. MT5 doesn’t include it in its standard indicator set, so you need a community-written version.

A detail people often miss: a breakout signal has to be compared with the channel of the previous bars, not including the current one. If the current bar is included, the upper line always equals that bar’s own High, and price can never be above it. On TradingView the line does include the current bar, so check signals against the previous bar’s value.

Worked example: a 5-bar channel

N = 5 keeps the math simple. Example gold prices, not real quotes.

Bar	High	Low
1	2,410	2,398
2	2,415	2,402
3	2,412	2,400
4	2,418	2,405
5	2,414	2,401

1. Upper = highest High = 2,418 (bar 4)
2. Lower = lowest Low = 2,398 (bar 1)
3. Middle = (2,418 + 2,398) ÷ 2 = 2,408
4. Width = 2,418 − 2,398 = 20, about 0.83% of the midline

If bar 6 trades up to 2,421, it breaks 2,418, an upside breakout of the 5-bar channel. When bar 7 arrives, bar 1 drops out of the window, so the lower line moves up to the lowest Low of bars 2 to 6. That is why the lines move in steps instead of curving like a moving average.

How to read the channel

• Price at the upper line = a new high for the last N bars, buyers in control. The lower line is the mirror image
• Upper line stepping up, lower line following slowly = an uptrend
• Both lines flat, price moving inside = a range, nobody winning yet
• The midline shows whether price is in the upper or lower half of the recent range. Some traders use it as an exit or to move the Stop loss
• Channel width is a simple volatility gauge. Narrow = a quiet market, wide = price running hard. A narrow channel doesn’t tell you which way price will leave

In the chart, the channel is only 4.33 wide during the range. Once the trend starts it widens to 19.01, about 4.4 times as much. So the same number tells you both that the market is quiet and how wide a “normal” Stop loss distance is right now.

The Turtle rules, as they are usually told

In 1983 Richard Dennis and William Eckhardt taught a group of traders, the Turtles, to trade by clearly written rules. The rules that later became public were built on the Donchian Channel. The best-known version:

Part	System 1	System 2
Entry	Break of the 20-day High or Low	Break of the 55-day High or Low
Exit	Break of the opposite 10-day line	Break of the opposite 20-day line
Stop loss	2N from entry	2N from entry

N here is a smoothed 20-day average of True Range, which is the same as a Wilder 20 ATR. Position size was set so that a 1N move hit the account by about 1%. We tell this as history. It is not a claim that the rules work in today’s markets. Markets, costs and price behaviour have changed a lot since then.

What is worth noticing is the shape: there is no profit target. A trade is held until price breaks the shorter opposite line, so the exit trails the trend by itself, as in the chart above, which entered at 103.36 and exited at 120.00. Trend-following systems live on a few trades that run very far. Cap profits at a near target and those trades disappear. More on this in Near or far Take profit and Ride the trend or fade it.

False breakouts, and why breakouts need wide stops

A false breakout is when price pokes just outside the channel and then comes back in. In a range this happens a lot, because the channel lines are where orders wait: Buy stops from traders waiting for the breakout and Stop losses from traders holding the other side.

A Stop loss right next to the channel, for example just under the upper line, gets hit by a normal retest. So stops are usually set by volatility instead, for example 2 × ATR, which adjusts to the market by itself: shorter in quiet markets, longer in fast ones. More on ATR in ATR for volatility and Stop loss.

Example: buy at 2,418 on the upper break, ATR 20 = 9.5.

1. Stop loss = 2,418 − 2 × 9.5 = 2,399, a distance of 19 dollars
2. A 3,000-dollar account risking 1% = 30 dollars
3. On gold at 0.01 lot, a 1-dollar price move = 1 dollar, so 19 dollars of distance costs 19 dollars per 0.01 lot
4. 30 ÷ 19 = 1.58 times 0.01 lot, rounded down to 0.01 lot. The actual risk is 19 dollars, or 0.63% of the account

A wide Stop loss doesn’t mean more risk if the position shrinks to match. The full method is in Position size from your Stop loss.

Most breakouts don’t follow through. A breakout system survives on a Stop loss that is wide enough, a position that is small enough and a few trades that run far, not on guessing which breakout is real.

Our tests: an 8-bar box breakout system on gold

In our backtest, we took apart an 8-bar box breakout system on gold M1. It enters when price breaks the High or Low of the previous 8 bars, with extra volatility and momentum filters. Its Stop loss is wide, about 0.6% of price (about 26 dollars in 2026), but it takes small profits, starting at about 1.3 dollars. All figures are price distance.

Finding 1: the high win rate comes from the shape, not the entry. On real 2026 tick data, the system won 92.8–95.8% across several test variants. That sounds great. But random entries, counting whether price touches +1.3 before −26, win 95.3%, about the same. So the breakout entry doesn’t pick direction better than chance. The only part better than random is the average winning trade, 2.3 dollars against 1.3 for random entries, and that edge is thin: in the weakest test year, spread ate about 55% of the profit per trade before costs. The math is in A high win rate doesn’t mean profit.

Finding 2: price almost always comes back. 82% of winning trades pulled back past the entry before reaching profit. Only 18% broke out and kept going. That matches the false breakout picture above, and it is why every way of tightening the Stop loss in that test set lowered the win rate along with it.

Finding 3: the profit sits in one year. Tested with simulated ticks from 2022 to early September 2026, at 0.01 lot.

2026 alone made 3,460.68 of the 5,530.67-dollar total, 63% of the profit from 25% of the trades (952 of 3,766). And simulated ticks make results look better than they are. With the same settings, 2026 made +3,460.68 on simulated ticks but only +915.30 on real ticks, a 3.78× difference. A system that looks positive every year may only work while gold is very volatile.

All of this is backtest results. Past results do not predict future results, and they are not returns you should expect.

When the Donchian Channel misleads

• Ranging markets. Price hits the upper and lower lines in turn, and every breakout is a false one. A common filter is trend strength, for example ADX, but every filter also throws away some good trades
• N too short on a small timeframe. The channel is so narrow that spread and noise break it every minute, and cost per trade eats the profit. See Spread cost on small timeframes
• Big news. Price breaks out and comes back within 1–2 bars, and a Buy stop may fill well past the line
• After a long trend. The channel is very wide, so an ATR stop is wide too, and the position must shrink further

If you pair it with other indicators, choose ones that measure something different. Donchian and Bollinger Bands both measure how wide price is ranging, so using both adds little. See the approach in Combining indicators without redundancy.

Every indicator describes past price. Donchian only tells you price has made a new high or low. It doesn’t tell you price will keep going.

Common mistakes

• Comparing the signal with a channel that includes the current bar, then wondering why price never breaks the line, or getting backtest signals that use future data
• Putting the Stop loss right next to the channel line and getting swept out by a normal retest. Use an ATR distance and a smaller position instead
• Adding a near profit target to a breakout system. The win rate rises, but it cuts off the trades that run far, which are the ones that actually pay
• Seeing a high win rate and believing the entry is good. Compare it first with random entries at the same TP and SL distances
• Quitting after a losing streak. Long losing streaks are normal for breakout systems, see Losing streaks are normal. But don’t trust a backtest whose profit sits in a single year either
• Tuning N to fit past charts. The best value in the past is rarely the best in the future