How to Use Donchian Channels
To use Donchian channels, take a break of the upper or lower band as an entry in that direction and use a shorter channel in the opposite direction as the exit. The bands are literally the highest high and lowest low over the lookback, with nothing else in the calculation.
Donchian channels plot the highest high and the lowest low over a lookback period. That is the whole calculation. It is the most transparent system in common use and the one whose failures are easiest to attribute.
Before you start
A lookback period chosen from your holding horizon rather than from a default. Twenty bars and fifty-five bars behave very differently and both are defensible.
An acceptance that most breakouts of this kind do not continue. On this site’s shared series price traded through 85% of 39 twenty-bar levels and 100% of 11 fifty-five-bar levels — reaching a level is ordinary.
A separate, shorter channel decided in advance as the exit. Entering on a long channel and exiting on a short one is the actual system, not an optional refinement.
The steps
1. Understand what the bands are
The upper band is the highest high of the last N bars; the lower is the lowest low. No averaging, no weighting, nothing that could be tuned.
2. Choose the entry lookback from your horizon
A short lookback signals often and early; a long one signals rarely and late. Neither is better, and the choice should follow how long you intend to hold.
3. Take the break as the entry
Price closing beyond the band is the trigger. Intrabar breaks reverse frequently, so the close is the version that does not change its mind after you have acted.
4. Use a shorter channel as the exit
Enter on a fifty-five-bar break, exit on a twenty-bar break the other way. This asymmetry is what keeps losing trades short and lets winning ones run, and it is the whole design.
5. Expect most entries to fail
The system accepts a low proportion of winners in exchange for the size of the ones that work. Judging it on how often it is right is judging it against a design it never claimed.
6. Size from the exit distance
The distance from entry to the opposite short channel is what you are risking. That distance divided into your risk figure gives the size, and it changes with volatility.
7. Take every signal or none
The infrequent large winner is what makes the arithmetic work. Skipping signals that look unconvincing removes them at exactly the rate they occur, because they always look unconvincing.
How to tell it worked
The entry and exit lookbacks were both chosen before the first trade.
Every entry was on a bar close, so 0 came from an intrabar break.
Position size came from the entry-to-exit distance rather than from a fixed number of units.
And every signal in the last 90 days was taken, none of them skipped for looking weak.
What it does not contain
Any view about quality. A break driven by real participation and one driven by a single order look identical, because the calculation only sees the highest high.
Any protection against a narrow range. In a quiet market the channel is tight and price exceeds it constantly, producing a stream of signals about almost nothing.
Why the asymmetry matters
A long entry channel means you only act on substantial moves. Fifty-five bars is roughly a quarter of daily data, so exceeding it means price is at a level it has not seen in months.
A short exit channel means you leave quickly when it fails. Twenty bars in the other direction cuts a losing trade before it becomes a position you are managing rather than trading.
Reversing the two produces the opposite system and it does not work. Entering easily and exiting reluctantly is the natural human configuration, which is precisely why the rule has to be written down before any of it starts.
Choosing the two lookbacks
The classic pairing is a long entry channel and an exit channel about a third of its length. That ratio is a starting point rather than a discovery, and its virtue is that it was chosen before anyone looked at your instrument.
A shorter entry channel trades far more often. More signals, more round trips, and a larger share of them are breaks of a range rather than the start of anything.
A longer exit channel holds losing trades longer. It also holds winning ones longer, and which effect dominates depends on how often large moves actually occur in what you trade.
Change one at a time and only with a reason. Adjusting both after a losing run produces a configuration fitted to that run, and the next one will look different in a way the new settings do not anticipate.
The original data
Of the 24,971 unique videos in research/search-study-corpus.jsonl, 59 mention these channels in the
title, at a median of 10,071 views across 54 channels, and 64% of those titles are instruction-shaped.
One neighbouring band system appears in 126 at 3,163 and a trend line indicator in 122 at 19,638. The
counts come from site/corpus_count.py.
59 videos at 10,071, the thinnest coverage of the channel indicators measured here. The simplest system attracts the least explanation, which is a consistent pattern — there is very little to demonstrate when the calculation is a maximum and a minimum.
The answer to the question on that chart is that skipping is how this system stops working. A minority of signals produces the result, they are indistinguishable in advance, and a run of four failures is the ordinary texture of an approach that accepts a low proportion of winners by design.
When it fails
The failure is discretion applied to a mechanical system, and it removes exactly the trades that matter. After a run of losses the next signal looks unconvincing, so it gets skipped. It runs. The one after that gets taken and fails. Over a year the trades that were skipped are disproportionately the large winners, because a large winner starts out looking exactly like a failed breakout and only distinguishes itself later.
The second failure is judging it on win rate. It is designed to be wrong often.
A third is symmetric channels. Entry and exit should not be the same length.
A fourth is acting intrabar. Breaks reverse before the close constantly.
A fifth is a fixed position size. The exit distance moves with volatility.
And a sixth is using it in a range. A tight channel signals on nothing.
Related
Donchian channels covers the calculation. Breakout is the event the system trades and how often it continues. And trend following is the approach the asymmetry belongs to.
What makes this worth understanding is that it has nowhere to hide. There are no weights, no smoothing constants, no source options. If it works it works because breakouts of a recent range sometimes continue, and if it fails you know exactly which assumption failed — which is more than most indicators offer.
— Michael Whitman
This page is educational, not financial advice. Test every idea on your own charts before risking money.