Bollinger Bands vs Keltner Channels

Michael Whitman · Trading since 2017, full-time since 2020 · Published 2026-09-08 · Updated 2026-10-04 · Reviewed 2026-10-04

Both draw an envelope around an average of price. Both are used to judge whether price has stretched. The difference is in how the distance from the middle is measured, and on thirty-two years of SPY daily bars that one choice changed how often price closed outside, which side it closed on, and how violently the width moved.

What each one is

Bollinger Bands set the distance with standard deviation. The default is a 20-day simple average of closes with lines two standard deviations of those same 20 closes above and below. John Bollinger’s own site says he uses the population calculation, dividing by 20 rather than 19; the sample form would make every band 2.60% wider. Bollinger Bands covers the calculation.

Keltner channels set the distance with average true range. Chester Keltner’s 1960 original used a 10-day simple average of the typical price, (high + low + close) ÷ 3, with lines one 10-day average high-to-low range away. The version most charting packages show is the one Linda Bradford Raschke introduced in the 1980s, which sets the distance with average true range; StockCharts’ defaults for it, used here, are a 20-day exponential average of closes with lines two 10-day average true ranges away. This page measures that version unless it says otherwise. Keltner channels covers both.

The middle line is close in both. A 20-day simple average and a 20-day exponential one of the same closes, which is why the envelopes share their lag after a sharp move.

Where they differ

What the width reads. Standard deviation reads how spread out the last 20 closes were. Average true range reads how large the bars themselves were, including any gap from the previous close, so one sees overnight movement directly and the other sees only as much of it as survives into the close.

How the width responds to a shock. One large close moves a standard deviation sharply, because that is what an outlier does to dispersion. A Wilder average of true range absorbs a big bar a tenth at a time, so the Keltner width moves in smaller steps.

Which side price closes outside. Bollinger’s outside closes split almost evenly above and below. Keltner’s leaned heavily to the upside on SPY, so on this index a close above its upper line was common and a close below its lower line was not.

How often price touches an edge. On SPY, Keltner’s lines were touched on more sessions than Bollinger’s, not fewer. Neither envelope makes a touch rare.

How the width measures pull apart

Standard deviation collapses in a steady trend. When closes march up by similar steps, they are not very spread out, so the Bollinger width shrinks even while bars stay a normal size. Average true range keeps measuring the bars, so the Keltner width does not shrink with it. That is the squeeze: the Bollinger lines falling inside the Keltner lines.

And it explodes on a single close. A shock day enters the standard deviation at full size and stays for 20 sessions, then drops out all at once.

A worked example from 2026. On 4 August 2026, SPY’s Bollinger lines sat at $764.60 and $729.79, $34.81 apart, while the Keltner lines sat at $767.14 and $726.98, $40.16 apart: the Bollinger lines were inside, the tenth straight squeeze session since 22 July. On 5 August the lower Bollinger line, $728.45, dropped below the lower Keltner line, $729.68, and the squeeze ended. Over the next 20 sessions SPY’s highest high and lowest low were only 2.58% of the 5 August close apart, less than half the median for any 20 sessions in the data below, rather than the expansion the squeeze is often said to precede.

Raschke’s channel against Keltner’s 1960 original

The name covers two different envelopes. Keltner’s own rule centered a 10-day average of the typical price and set each line one average daily range away, with no gap from the previous close counted. The newer version, at its usual defaults, swaps in a 20-day exponential average and two average true ranges, which does count gaps.

The original is much tighter. One average range either side of a 10-day line leaves price outside it far more often than either modern envelope, as the table in the data section shows. A squeeze reading built on the original would almost never fire, because the Bollinger lines are rarely inside something that narrow.

So the version has to be named. A rule written on one platform’s Keltner channel and tested on another’s can be testing a different envelope without anyone noticing.

Which one to use: Bollinger Bands or Keltner Channels?

Run Keltner channels when you want a width that changes gradually. On SPY its day-to-day width changes were smaller, and big one-day jumps far rarer, as the data section shows.

Run Bollinger Bands when you want the width to react to a shock immediately. The standard deviation widens on the close that surprised, which is exactly what it is built to show.

Run the range version if you also size positions from average true range, because then the envelope and the stop distance come from the same measure.

And run one rather than both when the plan is a squeeze reading. The squeeze is a comparison between the two, and on SPY it was on for more than a quarter of all sessions.

The original data

The sample. Yahoo Finance daily SPY bars, highs, lows and closes as traded, from 29 January 1993 to 2 October 2026; measured from 3 January 1994, so every line has a full year to settle: 8,243 sessions. Opens are not used. Bollinger: 20-day simple average ± 2 population standard deviations. Keltner: 20-day exponential average ± 2 × Wilder’s 10-day average true range. A close outside means above the upper line or below the lower; a touch means the high reached the upper line or the low reached the lower.

In sessions, that is 442 closes above and 439 below the Bollinger lines, and 1,076 above and 350 below the Keltner lines.

They disagreed on one session in seven. On 1,217 sessions (14.76%) the close was outside one envelope and inside the other: 881 times outside Keltner only and 336 times outside Bollinger only. The longest run of closes above the upper line was 9 sessions for Bollinger and 25 for Keltner, the latter ending on 15 May 2026. Keltner’s 1960 original is far narrower than either, and price closed outside it on nearly half of all sessions.

The widths moved differently. The Bollinger width changed by a median 4.30% a day and the Keltner width by 2.87%; the Bollinger width jumped 10% or more in a day on 793 sessions against 314 for Keltner, and its largest one-day widening was 120.04%, on 9 September 2016, against 57.62%. The width ratio moved a long way. The Bollinger width divided by the Keltner width had a median of 1.08, but a tenth of sessions were below 0.71 and a tenth above 1.71.

The squeeze. Counting a squeeze day as one where both Bollinger lines sat inside both Keltner lines, SPY spent 2,346 of 8,243 sessions (28.46%) in one, across 298 completed spells (the first already running when the measurement began) with a median length of 6 sessions and a longest of 40. With the Keltner multiplier cut to 1.5, the count fell to 466 sessions (5.65%): how often a squeeze happens is set mostly by that setting. Against Keltner’s 1960 original, the Bollinger lines sat inside on only 19 sessions (0.23%).

What followed a release was ordinary. For each of the 297 releases with 20 sessions still to come, the test took the distance between SPY’s highest high and lowest low over the next 20 sessions as a share of the release-day close. The median was 5.90%, against 5.79% after any of the 8,223 sessions with 20 still to come; 159 of the 297 releases ranged more than the all-day median, p = 0.25 against one half. The quietest fifth of sessions (1,629 with 20 sessions to come) by Bollinger width were followed by a median range of only 4.24%, so quiet tended to be followed by quiet, while a release was followed by an ordinary range. Every session’s lines are in the SPY Bollinger and Keltner file, and every spell in the squeeze file.

The video record. Of the 24,971 videos in this site’s search study, 15 name both in the title, at a median of 4,829 views.

When it fails

The failure is fading a touch of either edge in a trend, and the envelope encourages it every bar. Price reaches the upper line, which reads as stretched, so a short is taken. The move continues and price rides the edge, which on SPY lasted 25 straight closes above the Keltner line in 2026. Each new bar along the edge makes the short look better reasoned while it loses more.

The second failure is treating a squeeze as a promise of movement. On SPY the 20 sessions after a release ranged about as far as any 20 sessions did, and it says nothing about direction either.

A third is running both for confirmation. On one session in seven they disagree, and neither says which one is right.

A fourth is placing a stop at an edge. Everyone can see it, and on SPY price touched one of the Keltner lines on more than a quarter of sessions.

A fifth is comparing settings across charts. A Keltner channel on one platform may be the 1960 version and on another the ATR version, and the two put price outside on very different shares of days.

And a sixth is optimizing the multiplier after a losing run. That is fitting the past.

Related

Bollinger Bands covers the standard-deviation envelope. Keltner channels covers the average-range one. And average true range covers the width measure underneath the second.

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