How it works
Keltner Channels wrap a moving average in an envelope whose width comes from average true range. The result looks like a Bollinger envelope but behaves differently in a way that matters. Standard deviation is a squared measure, so one violent bar can widen a Bollinger envelope enormously; average true range is a plain average of ranges smoothed slowly, so the same bar barely moves it. Keltner therefore produces a steadier channel that keeps its shape through shocks, at the cost of responding more slowly to a genuine regime change.
That steadiness makes the channel better suited to trend work than to reversion work. Because the width does not balloon on the first impulsive bar, price that breaks the upper channel tends to stay outside it for the duration of a real move, which turns a channel break into a usable continuation signal rather than the exhaustion signal traders instinctively read into it. Many systematic trend followers use exactly this: enter on a close outside the channel, exit when price closes back through the central average.
The centre line is normally an exponential moving average, which adds a second layer of usefulness. The slope of that centre is a trend read on its own, and the position of price relative to it defines the two halves of the channel. A common workflow is to take the centre as the bias, treat the far channel as a stretch marker, and use pullbacks toward the centre in the direction of its slope as entry locations. That approach is far more robust than fading the outer channel, which fails badly in exactly the markets where the channel is most useful.
The relationship with the Bollinger envelope is close enough that the comparison itself became an indicator. When Bollinger Bands contract entirely inside the Keltner Channel, dispersion of closes has fallen below typical true range, which is a precise and objective definition of compression, and the release of that state is the well-known volatility squeeze. Against Donchian channels, Keltner is smoother and centred, where Donchian tracks raw extremes and has no centre at all.
Calculation
The arithmetic in words, in the order it happens.
Compute the centre line as an exponential moving average of the close, conventionally over 20 periods. Separately compute average true range, commonly over 10 periods with Wilder smoothing, where true range for each bar is the largest of the high minus the low, the absolute difference between the high and the previous close, and the absolute difference between the low and the previous close. The upper channel is the centre plus a multiplier, conventionally 2, times ATR; the lower channel is the centre minus the same amount. The original 1960 formulation instead used a ten-day simple moving average of typical price for the centre and a ten-day moving average of the high-minus-low range for the width; the average-true-range version described here is the one every modern platform ships.
Source
An AlgoBeamScript implementation of the formula above, written by us from the arithmetic so the code and the calculation agree line for line.
Runs unchanged on the platform and in the AlgoBeamTS runtime. The language reference is in the documentation.
Inputs
Defaults are the values most charting packages ship with. They are conventions, not optimal settings — the right length depends on your instrument and your holding period.
| Input | Default | What it changes |
|---|---|---|
| Length | 20 | Bars in the centre-line average. Shorter values make the channel track price closely and produce frequent breaks; longer values give a stable trend spine that price can spend weeks either side of. |
| Multiplier | 2.0 | How many ATR the channel sits from the centre. Around 1.5 makes breaks common and suits intraday continuation work; 2.5 to 3 reserves breaks for genuinely impulsive moves and is the usual choice on daily bars. |
| ATR length | 10 | Bars in the true range average that sets the width. Shorter values let the channel react quickly to a volatility change; longer values keep the width stable through a shock and are preferable when the channel drives stops. |
| Average type | EMA | Smoothing used for the centre. An exponential average responds faster to a turn; a simple average is steadier and keeps the channel closer to the original design. |
How to read it
What practitioners take from the plot. Read these as descriptions of market state, not as entry signals.
- Close outside the upper channel
- Price has moved more than a typical multiple of true range beyond its own average. In a trending market this is continuation, not exhaustion.
- Price riding between centre and upper channel
- A healthy, orderly uptrend. Pullbacks that hold the centre line are the standard continuation entry for channel systems.
- Close back through the centre after a run
- The move has lost its slope. Trend-following exits are usually placed here rather than at the opposite channel.
- Channel width shrinking with a flat centre
- True range is contracting around a directionless average, the compression state that precedes an expansion.
- Bollinger Bands entirely inside the channel
- Dispersion of closes has fallen below typical true range, the strict definition of a squeeze and the setup for a volatility breakout.
Limitations
Where this indicator misleads. None of these are fixed by a better parameter.
- The channel is directionless about breaks: an upper break in a range is a fade and an upper break in a trend is an entry, and the channel cannot tell you which regime you are in.
- ATR responds slowly, so after a sudden volatility collapse the channel stays too wide for a while and breaks stop firing when they would be most useful.
- Because the centre is a moving average, everything the channel says lags. In a sharp reversal price can cross the entire channel before the centre has begun to turn.
- Defaults vary widely between platforms, with different ATR lengths, multipliers and average types all shipping as standard, so a rule copied from one chart may behave very differently on another.
Educational reference. This page explains how an indicator is built and how it is commonly read. It is not investment advice, not a recommendation and not a signal service. No indicator is profitable on its own — each is a way of describing a market, and any rule built on one has to be tested with realistic costs before it is traded.