Home Features Multi-market scanning Alerts Backtesting Entry check Parameter search AI strategy builder Refine Plans Guides FAQ About Careers Contact Log in
GUIDE

Keltner Channels explained, and how to write them as a condition

Volatility · 4 min read

Keltner Channels are a moving average with a band above and below it, each a fixed number of Average True Ranges away. They look like Bollinger Bands and answer a similar question, with one difference in how the width is set that changes how they behave.

Keltner Channels around candles — an EMA with bands two ATRs above and below — with the bar where price closes above the upper band highlighted
Illustrative candles. The dashed line is the EMA; the bands sit two ATRs either side and widen as the bars do.

What they measure

The middle line is an exponential moving average of the close, twenty bars by default. The upper band is that average plus a multiple of the ATR — two by default — and the lower band is the same distance below. ATR is the average size of a bar, so the channel is "the average, plus or minus two typical bars' worth of movement".

Bollinger Bands set their width from the standard deviation of the closes, which jumps when price makes a sudden move and collapses when it goes quiet. ATR is an average of ranges and moves more gradually, so Keltner Channels keep a steadier width: they do not balloon on one big bar, and they do not pinch as tightly in a lull. That is the difference, and it is why some people draw both and read the Bollinger Bands going inside the Keltner Channels as a squeeze.

The settings

  1. Period. The length of the EMA in the middle and of the ATR. Twenty by default.
  2. ATR multiplier. How many ATRs away the bands sit. Two by default; 1.5 brings them in so price reaches them more often, 3 pushes them out so only an unusual move does.
  3. Touch tolerance. How close to the middle band a close has to be to count as reaching it, as a percentage of price. Only the middle-band condition uses it.

Writing them as a condition

These are the comparisons SixOne8 offers, in the builder's own words. The distinction between reaching and breaking is the one to get right.

  1. Price reaches the upper / lower band. The bar closed at or beyond the band. True on every bar that closes there, including several in a row — a state.
  2. Price breaks above / below. The previous bar closed inside the band and this one closed beyond it. True only on that bar — the moment, and the usual trigger.
  3. Price reaches the middle band. The close is within the touch tolerance of the EMA, from either side. The condition for "price has come back to the average".

Keltner(20, 2.0) on 4h  ·  price breaks above

Read it as a sentence: on the 4-hour chart, the close has just moved from inside the channel to above its upper band. That is what the scanner evaluates at each 4-hour close.

Two readings of the same band

A close beyond the upper band is read two opposite ways, and the app takes neither side. One strategy treats it as a stretch — price is two typical bars above its average, which is far — and looks for the move back towards the middle band. Another treats it as a breakout: the move has carried further than usual and may be the start of a trend. Which reading holds on the market and timeframe you care about is a question the entry check answers, and it answers it differently on different markets.

Before you trust them

Run an entry check. It will tell you whether, on the market you care about, the moments this condition fired moved any differently from a random moment over the history tested. That is a fact about the past on that market, and it is the only kind of fact a test here produces.

Next

Bollinger Bands explained The standard-deviation version, and the squeeze. ATR explained The unit the bands are measured in.