What Are Keltner Channels?
Keltner Channels are an exponential moving average with a channel drawn a multiple of ATR above and below it, typically 1.5 or 2. Because the width comes from average bar range rather than standard deviation, the channel reacts to volatility differently from Bollinger Bands.
Chester Keltner’s channel and John Bollinger’s bands look almost identical on a chart and are built from two different ideas about what “volatile” means.
How it works
Three lines. The middle one is a 20-bar exponential moving average.
That is already a difference from Bollinger Bands, whose basis is a simple average. An EMA weights recent bars more heavily, so the Keltner centre line turns sooner.
The band sits a multiple of ATR away. On the last bar of this chart the ATR is 0.26, and 1.5 × that is 0.39 — which is exactly the distance from the middle line to the upper one.
Where it differs from Bollinger
Both measure how far price usually travels. They measure it differently.
Standard deviation squares each deviation before averaging, so one violent bar dominates the result. ATR averages the true range without squaring, so the same bar counts once.
The practical consequence: after a shock, the Bollinger band widens sharply and the Keltner channel widens gently. For a few bars afterwards the two disagree about how volatile the market is — and both are right about the question they asked.
That disagreement is measurable and useful. When the Bollinger band falls entirely inside the Keltner channel, statistical spread has dropped below typical bar range, which is the squeeze that a whole indicator is built on.
What it is good at
The middle line is a usable trend reference. In an advance, pullbacks toward the EMA are where being wrong is cheap — an ordinary moving-average job done by a line that happens to have a channel attached.
The band is not a ceiling. Five closes above it here, and the move continued through all of them. That is the same warning the Bollinger page gives and it applies identically, because it follows from what a channel is rather than from how it was computed.
The settings
Three numbers: the EMA length, the ATR length, and the multiplier.
The multiplier decides how often anything happens. Measured here: 5 closes outside at 1.5 ATR, 2 at 2.5.
1.5 and 2 are both common defaults, which is worth knowing before you compare your chart with someone else’s — a “break of the Keltner channel” is not a single event that two people would agree on.
Why ATR width matches how you already think
Worth its own note, because it is the practical reason to prefer this one.
Every stop on this site is sized in ATR. The ATR page argues that a fixed percentage stop is too wide in a quiet market and too tight in a loud one, and that the fix is to measure distance in units of typical bar range.
A Keltner channel is that same unit, drawn. The upper band is literally “1.5 stop-distances above the average”, so the channel and your risk framework are speaking the same language.
A Bollinger band is not. It is two standard deviations, which is a statement about the distribution of closes rather than about the size of bars — a perfectly good measurement in different units.
Neither is more correct. But if your position sizing is ATR-based, one of these two channels is already commensurate with it and the other needs translating.
A worked example
Read the middle line for direction. Rising EMA, price above it: the ordinary trend condition.
Use pullbacks to the basis, not touches of the band. The band is where price has stretched; the basis is where a trend that is still intact tends to find support.
Take the invalidation from price, below the last swing low — not from the lower band, which moves every bar.
And if the Bollinger band is inside the channel, wait. That is the squeeze, and it is a statement that nothing is happening yet.
The original data
Across our study of 24,971 trading videos, 127 cover Keltner Channels. The median one gets 2,967 views, 90% never pass 50,000, and the median length is 8.2 minutes.
That 90% is among the highest saturations measured here, close to Chaikin Money Flow at 96% — a small field with very little reach.
The corpus carries description text for 55 of those 127, and across those 55, two mention invalidation, failure, or what a bad read looks like.
When it fails
Sideways, price crosses the middle constantly
In a trading range the EMA sits in the middle of the noise, so price crosses it on nothing, and the channel edges are reached in both directions without either meaning anything.
The band is treated as a boundary
Covered above and it is the expensive one. Five closes outside during the advance — a channel describes where price has been, not where it may go.
It is running next to Bollinger for no reason
If you cannot say what the disagreement between the two means, you have one measurement twice. The squeeze is the one honest reason to have both.
The version you are running may not be Keltner’s
Chester Keltner’s original, from 1960, used a simple average of the typical price and a band set by the average daily range — not an EMA, and not ATR. ATR did not exist yet; Wilder published it in 1978.
What almost every platform now ships is Linda Raschke’s revision, which swapped in the exponential average and the ATR width. That is the version described on this page, because it is the one you will actually load.
So “Keltner Channel” names two different calculations, and the older one still appears in some software. Check what your platform’s settings panel offers — if it asks for an ATR length, you have the modern one; if it asks only for a length and a multiplier, you may not.
You read the squeeze once it had resolved
A narrow channel that preceded a rally and one that preceded a fall look the same, because width has no direction in it.
Related
Bollinger Bands is the other channel, and the comparison above is the reason to know either.
ATR is the width, and the page that explains why measuring in bar ranges is worth doing.
And squeeze momentum is the indicator built entirely out of these two channels disagreeing.
I ran both these and Bollinger for a while before realising I was looking at one question twice, and then I kept Keltner because the ATR width matches how I already think about stop distance. The genuinely useful thing I did not expect is that when the two channels disagree it means something specific, and that is the only reason to have both on a chart at once.
— Michael Whitman, from this video
This page is educational, not financial advice. Test every idea on your own charts before risking money.