Donchian Channels are price channels built from recent extremes. They plot an upper band at the highest high of the last N bars, a lower band at the lowest low of the last N bars, and often a middle line that is simply the average of the two bands. The purpose is not to forecast direction, but to show where price is relative to its own recent range in a rules based way.
What they measure is range expansion and range containment. When price pushes above the upper band, it is making a new N bar high and signaling that buyers are willing to pay more than they have in the recent window. When price falls below the lower band, it is making a new N bar low and signaling that sellers are willing to accept less than they have in the recent window. In that sense, the channel is a clean breakout detector that stays anchored to the market you are trading rather than to a fixed percentage.
Donchian Channels also provide structure for trend following. In sustained trends, price often rides the outer band, and pullbacks tend to stay inside the channel rather than breaking the opposite side. That makes the channel useful for both entries and exits, especially if you want a simple framework that does not depend on many parameters.
How Donchian Channels are calculated
The calculation is straightforward and works on any timeframe because it only needs highs and lows. Choose a lookback length N, such as 20 bars on a daily chart or 55 bars on a weekly chart. Then compute the highest high and lowest low over that rolling window.
A simple formula version looks like this:
Inputs
- N = number of bars to look back, example 20
- High = the highest price of each bar
- Low = the lowest price of each bar
Upper Band
Upper Band is the highest High from the last N bars.
- Upper Band = highest(High, N)
Lower Band
Lower Band is the lowest Low from the last N bars.
- Lower Band = lowest(Low, N)
Middle Band
Middle Band is the midpoint between the two bands.
- Middle Band = (Upper Band + Lower Band) / 2
Example
If N = 20, you are always looking at only the last 20 bars:
- Upper Band is the highest high in those 20 bars
- Lower Band is the lowest low in those 20 bars
- Middle Band is the average of those two numbers
Two implementation details matter in practice. First is whether the breakout uses the current bar or the prior completed bar for the channel values, because that changes repaint like behavior during an active candle. Second is whether your entry rule triggers on intrabar touches or only on a bar close beyond the band, which changes the trade frequency and the false breakout rate.
Most used settings and why traders choose them
The most common Donchian Channel setting you will see is 20 bars. It is popular because it fits roughly one trading month on daily charts and it captures many intermediate breakouts without being too fast. Traders who like swing breakouts often start at 20 and then adjust based on the instrument’s noise level and the holding period they can tolerate.
A second widely used setting is 55 bars, associated with longer term trend following and the classic Turtle style breakout logic. The idea is that a longer window reduces churn and focuses on moves that are significant relative to the last few months. The tradeoff is later entries and wider exits, which can be fine if you are aiming to ride large trends rather than catch every smaller push.
Shorter settings like 10 or 14 bars are used when traders want earlier signals. They can work better on markets that trend smoothly or on lower timeframes where you need faster adaptation. The downside is frequent band breaks during sideways conditions, which creates whipsaws unless you add filters like trend strength, volatility regime checks, or a time based rule for re entry. Many traders end up with a two layer approach, a shorter channel for entries and a longer channel for exits, so the system can get in earlier but stay in longer when trends develop.
How Donchian Channels behave on charts
On a chart, Donchian Channels expand and contract as the lookback window rolls forward. During consolidation, the upper band and lower band flatten because new highs and lows are not being set, and the channel becomes a visual map of a range. When price breaks out, the outer band steps in the breakout direction as new extremes are made, and the channel starts sloping if the trend persists.
A common behavior in strong uptrends is band riding. Price will frequently tag or stay near the upper band, and pullbacks often remain well above the lower band. In downtrends, the mirror happens near the lower band. This is useful because it helps you avoid interpreting every pullback as a reversal, especially if the opposite band is far away and not being threatened.
Another important behavior is that Donchian Channels do not normalize for volatility the way ATR bands do. A narrow channel can occur because the market is quiet, and a wide channel can occur because the market is active, but the breakout threshold is always the recent extreme. This makes the indicator very direct, but it also means you must be careful about breakout quality. A small poke above the upper band in a low volatility range can be meaningful, while a similar poke during high volatility can be noise.
When Donchian Channels tend to work and why
Donchian Channels tend to work best in markets that show directional persistence. That includes trending equities after strong catalysts, futures markets during sustained macro moves, and any instrument where breakouts are followed by continued participation. The reason is simple: the method assumes that new highs and lows can lead to further movement in the same direction, at least often enough to offset losses from failed attempts.
They also work better when the market transitions from compression to expansion. If price has been contained and the channel is relatively flat, a decisive break can mark a real change in order flow. In those moments, the channel is good at turning an abstract idea like breakout into a mechanical rule, because the level is defined and not subjective.
Regime filters often improve results because they reduce trading during chop. Trend strength measures are one way to do this. For example, you can require a minimum trend strength reading before taking breakouts, or use the concept explained in How to use ADX Average Directional Index to separate weak drift from stronger directional phases. Another approach is simple time based logic, such as only trading breakouts that occur after a minimum number of bars inside the channel, which helps avoid overtrading in noisy ranges.
When Donchian Channels tend to fail and why
The most common failure mode is whipsaw in mean reverting markets. When price oscillates around a value area, it can repeatedly set marginal new highs and lows without follow through. Because the channel is always based on recent extremes, it will trigger signals even when the move has no real expansion behind it. This is especially visible when the channel is flat and price repeatedly pokes beyond the band and snaps back.
Another failure mode is late entries in mature trends. A long lookback channel can keep you out until a trend is well advanced, and then the first entry happens near an exhaustion phase where the risk reward is less favorable. This is not a flaw of the indicator, it is a consequence of choosing a slow window. If you want early entries, you need either a shorter channel, a pullback entry rule, or a layered system that uses the channel as confirmation rather than as the first trigger.
Gap risk is also important, especially in single stocks. A breakout can occur via a gap above the upper band, which can be a strong signal but also compress your stop placement options. If you use an opposite band exit, the distance can become very large after a gap. If you use a tighter stop, you can be stopped out by normal volatility even if the trend later continues. This is why many traders pair Donchian logic with volatility based position sizing and a clear rule for where the trade is invalidated.
Practical rules for entries, exits, stops, and filters
A clean baseline entry rule is to buy on a daily close above the upper band for a long trade and sell short on a daily close below the lower band for a short trade. Using the close reduces noise compared with intraday touches and makes backtesting more consistent across data sources. If you trade intraday, you can still use the same logic but apply it on your chosen bar interval and stick to close based confirmation.
For exits, there are two common approaches. One is a channel exit, where a long trade exits on a close below the lower band, and a short trade exits on a close above the upper band. This keeps the system symmetrical and lets trends run, but it can give back a meaningful portion of gains. The second approach is a trailing stop that sits inside the channel, such as a recent swing low for longs or swing high for shorts, which can reduce giveback but increases the chance of getting shaken out early.
Stops should be tied to the reason you entered. If you enter because price broke the upper band, then the first sign the breakout failed is often a close back inside the channel and continued weakness. A practical stop is below the breakout level or below a recent structural low, but the exact placement should align with volatility and your timeframe. If you can only tolerate tight stops, you may be better using Donchian Channels as a trend filter and using a different trigger for entries, because tight stops and breakout systems can conflict.
Keep filters simple and connected to known failure modes. A trend filter can be slope based, such as requiring the middle band to be rising for long trades, or requiring price to be above a longer moving average. A volatility filter can require that the channel width is above a minimum threshold, which avoids trading tiny ranges that produce frequent fake breaks. A momentum filter can be as simple as requiring that the breakout bar closes in its upper portion for longs and lower portion for shorts, which reduces entries on weak closes.
One simple two condition filter that many traders find practical is to combine Donchian breakouts with a trailing trend line style indicator, so you avoid taking longs when the broader trend structure is down. If you already use an ATR based trend line, you can align the breakout with it, similar to the regime logic described in Supertrend Indicator Explained With Practical Trade Rules. The point is not to stack indicators, it is to reduce the specific kind of chop where Donchian signals are most vulnerable.
Here is a compact ruleset you can test without overfitting:
- Entry long: close above upper band, middle band rising, avoid entries if channel width is extremely narrow
- Entry short: close below lower band, middle band falling, avoid entries if channel width is extremely narrow
- Exit long: close below lower band or a trailing swing low stop, whichever comes first
- Exit short: close above upper band or a trailing swing high stop, whichever comes first
Position sizing should reflect stop distance. If you use opposite band exits, the stop distance can be large, so position size should be smaller. If you use a tighter structural stop, size can be larger, but accept a higher stopout rate. The goal is consistency: the system is expected to take losses, and your risk model must survive those sequences.
Summary
Donchian Channels are a price channel indicator built from the highest high and lowest low over a chosen lookback period. They measure where price is relative to its recent extremes and provide objective breakout levels without subjective drawing. The classic calculation is upper band equals highest high over N bars, lower band equals lowest low over N bars, with an optional middle band as their average.
Common settings include 20 bars for intermediate breakouts and 55 bars for longer term trend following, with shorter windows producing earlier but noisier signals. The indicator tends to work best in trending and expansion regimes where breakouts have follow through, and it tends to fail in mean reverting chop where marginal new highs and lows reverse quickly. Practical use improves when you add simple filters tied to the failure modes, such as trend strength checks, channel width thresholds, and close based confirmation.