Keltner Channels vs Donchian Channels

Michael Whitman · Trading since 2017, full-time since 2020 · Reviewed 2026-09-08

Two channels drawn from completely different logic. One steps to whatever extreme the market has made; the other flows with a measure of how much the market typically moves.

What each one is

Keltner channels place lines a multiple of average true range from a moving average. The width responds to how large bars have been. Keltner channels covers it.

Donchian channels plot the highest high and lowest low over a lookback. Nothing is smoothed; the lines sit at prices the market actually made. Donchian channels covers it.

One is a distance and the other is a level. That distinction runs through everything else on this page.

Where they differ

How the lines move. The Keltner envelope adjusts every bar. A Donchian edge is fixed until a new extreme is set, then it steps.

Whether the line is a real price. A Donchian edge is a price that traded. A Keltner edge is a computed offset from an average.

Who else can see it. The highest high of the last twenty bars is a number anybody can compute identically. Your Keltner multiplier is yours.

What each supports. Donchian supports a breakout rule with a recognisable event; Keltner supports a volatility-scaled envelope for judging stretch.

Where they agree

Both are backward-looking. Every value in each comes from closed bars, and neither anticipates anything about the next one.

Both need a lookback. The period decides responsiveness in both cases, and neither convention was chosen for your instrument.

Both fail in a range. On this site’s shared series direction runs average 2.01 bars with a longest of 11, and short runs produce constant touches and false breaks in either.

And neither supplies a stop. The ninetieth percentile bar range here is 1.101 and the largest single bar range was 2.338, which is what a stop at an edge has to survive.

Which one to use

Use Donchian channels for breakouts. A new extreme over a stated period is a checkable event, and on this site’s shared series 85% of 39 twenty-bar breakouts held.

Use Keltner channels when you want the width to adapt. Scaling by measured range keeps the envelope consistent when a market becomes more or less active.

Use Donchian when you want a shared level. Orders sit where lots of people can compute the same number, and a twenty-bar high qualifies where a private multiplier does not.

And use Keltner when you are judging stretch rather than breaks. How far price sits from its average relative to normal movement is the question that envelope answers.

Why stepping and flowing behave differently

Because a stepping level cannot creep away from you. It holds until the market makes a new extreme, which is what allows a break of it to be an event.

And because a flowing envelope adjusts to conditions. When bars grow, the Keltner width grows with them, which keeps the reading comparable across periods that a fixed level cannot.

What the measured breakout figures say

85% of 39 twenty-bar breakouts held on this site’s shared series, which is the base rate a Donchian rule works against.

100% of the 11 fifty-five-bar breakouts held, on eleven cases — suggestive rather than settled and worth reading as such.

Longer lookbacks produced fewer and better breaks. Waiting is the filter, and the cost of the filter is the trades you did not take.

And the underlying drift is mild. 54% of 566 ten-bar windows finished higher, so continuation starts from a modest tailwind rather than a coin flip.

What to set before using either

The lookback period. Twenty and fifty-five are the conventional Donchian settings; the Keltner length is a separate choice with its own effect.

The Keltner multiplier. It decides how far the envelope sits from the average, and on this site’s data average true range has a median of 0.5994 and a ninetieth percentile of 0.7954.

Whether a wick or a close counts. A wick beyond an edge and a close beyond it are different events.

And what the stop is. Neither channel provides one, and the largest single bar range here was 2.338.

The original data

Of the 24,971 unique videos in research/search-study-corpus.jsonl, no title compares these two directly — this pair is constructed from two subjects the corpus covers separately. Separately, Keltner channels appear in 119 titles at a median of 3,360 across 96 channels, and Donchian channels in 59 at a median of 10,071 across 54. The counts come from site/corpus_count.py.

119 videos on one at 3,360 and 59 on the other at 10,071. Twice the coverage and a third of the audience per video for the envelope — the plainer channel is taught half as often and watched three times as much.

The answer to the question on that chart is that they are reporting different things. One says an extreme was made; the other says price is not unusually far from its average — and both are correct.

When it fails

The failure is expecting the two to confirm each other, and they will disagree constantly by design. A new twenty-bar high can occur while price sits comfortably inside a volatility envelope, because a gently rising market makes new extremes without ever stretching far from its own average. Read as a conflict, that produces hesitation at exactly the points a breakout rule is meant to act on.

The second failure is treating a Keltner touch as an event. It is a distance.

A third is using a conventional lookback unchecked. It came from elsewhere.

A fourth is accepting a wick beyond an edge. A close is different.

A fifth is stopping at a channel edge. The largest bar range here was 2.338.

And a sixth is running both as one system. They answer separate questions.

Related

Keltner channels covers the range-scaled envelope. Donchian channels covers the extremes-based one. And average true range covers the measure the first is built from.

This page is educational, not financial advice. Test every idea on your own charts before risking money.