Bollinger Bands vs ATR
Michael Whitman
·
Trading since 2017, full-time since 2020
·
Published 2026-09-08
·
Updated
·
Reviewed
Bollinger Bands draw an envelope two standard deviations of closing prices either side of a 20-session average, ignoring wicks and gaps. Average true range measures the typical size of a whole bar including the gap from the previous close, which is what a stop distance has to survive.
Both get called volatility tools, and they are built from different parts of the bar. One reads where sessions closed; the other reads how far they traveled. Put the band into ATR units and the difference stops being abstract: the distance from price to the band changes all the time, while an ATR distance changes slowly.
What each one is
Bollinger Bands draw an envelope from the dispersion of closes. John Bollinger’s own site gives the defaults as a 20-period moving average with bands two standard deviations of the same 20 closes above and below, using the population form. The Bollinger Bands page walks through the construction.
Average true range reports how far a typical bar travels. Wilder’s true range is the largest of high minus low, high minus the previous close and the previous close minus low, smoothed over 14 bars. Average true range covers it.
One sees where bars finished; the other sees everything they did. Neither is a direction signal.
Where they differ
What enters the number. A gap between sessions goes straight into average true range. A band built from closes registers a gap only to the extent the close holds on to it.
Where each is anchored. The bands hang off a 20-session average, so the distance from today’s close down to the lower band depends on where the close sits inside the envelope. An ATR stop hangs off the close itself.
How each moves. Twenty closes can bunch together or spread apart within a week, which narrows or widens the bands quickly. Wilder’s average takes a fourteenth of each new bar, so it drifts rather than jumps.
What each is for. Judging stretch and reading squeezes on one side; stops and position sizes on the other.
What range and dispersion share
Both measure movement rather than direction, and both are built only from bars that have already printed. Each widens after a shock, not before it, and each needs a lookback that was not chosen for your instrument.
A gap day in September 2026
11 September 2026 shows the split. SPY had closed at $757.83 the session before, six cents above its lower band. It then traded entirely above that close, between $763.60 and $766.38, so its high-low range was $2.78 but its true range, measured from the previous close to the high, was $8.55. Average true range rose from $6.03 to $6.21. The 20-close standard deviation fell from $4.89 to $4.32, because the close of $764.29 landed back near the 20-session average of $766.88.
The two pictures disagreed. The bands had narrowed to $758.23 and $775.53, a half-width of $8.65, or 1.39 ATRs. The lower band sat $6.06 below the close, 0.97 ATR. A stop two ATRs down would have sat at $751.86.
Three sessions later both had been reached. On 16 September SPY traded down to $749.60, below the 11 September band value and below the two-ATR level, before closing between the two.
Which one to use: Bollinger Bands or ATR?
Use average true range when the number sets a stop or a size. A stop must outlast the whole bar, gap included, which is exactly what the measure reports, and it moves slowly enough to plan around.
Use Bollinger Bands when width itself is the reading. The squeeze is a genuine feature of a close-based envelope, and it has no direct equivalent in a range average.
Use average true range on anything that gaps. An instrument that opens away from the last close leaves a close-based measure blind to part of its movement.
And when both are on the chart, let one of them size the trade. Using the bands for the picture and ATR for the arithmetic is coherent. Placing stops at the band means a different risk on every trade.
How far below the close the lower band sits
It depends on where the close sits in the envelope. A close near the upper band leaves the lower band two half-widths away; a close near the average leaves it one; a close at the band leaves it nowhere.
And the half-width itself depends on how bunched the closes are. Measured in ATRs, it was below 1 on 60 sessions and above 4 on 267. A band that is a whisker away one week can be several ordinary bars away the next.
The original data
The sample. Yahoo Finance’s SPY daily highs, lows and closes, as traded, from 29 January 1993; every figure below covers 3 January 1994 to 2 October 2026. Opens are not used. Bollinger: 20-session average of closes ± 2 population standard deviations. ATR: Wilder’s 14-session version, seeded from February 1993.
The band’s width in ATR units. Half the band’s width (two standard deviations) divided by that day’s ATR had a median of 2.16. A tenth of sessions were below 1.42 and a tenth above 3.39. The narrowest was 0.61 ATR on 28 August 1995; the widest 6.28 on 24 November 2023.
Distance from the close down to the lower band. Median 2.53 ATRs, with a tenth of sessions under 0.382 and a tenth over 5.42. The lower band was nearer to the close than two ATRs on 3,237 sessions, 39.27% of the sample.
Which level a low reached within five sessions. For each close with five sessions after it (8,238), the test fixes two levels that day: the lower band’s value and the close minus two ATRs. Any low in the next five sessions at or below a level counts as reaching it. The band was reached on 2,132 sessions (25.88%) and the two-ATR level on 1,685 (20.45%). The overall figures hide the split:
| Starting day | Lower band reached | Close − 2 ATR reached |
|---|---|---|
| Band nearer than 2 ATR | 56.97% | 24.44% |
| 2 ATR nearer than the band | 290 days, under 6% | 17.88% |
By how wide the band was. Splitting the 8,238 days into fifths by half-width in ATRs, the band was reached within five sessions on 40.92%, 32.00%, 24.59%, 20.34% and 11.58% of days from narrowest to widest. The two-ATR level stayed between 16.55% and 22.89% in every fifth.
What this does and does not show. The band is not worse at anything; it simply measures a different distance, one that swings with how bunched the closes are. A stop distance set in ATRs kept a steadier chance of being reached. Every session’s band, ATR and both outcomes are in the SPY Bollinger width and ATR file.
When it fails
The failure is placing the stop at the lower band after a quiet stretch. The closes have bunched, so the band has crept up under price, sometimes to less than one ATR away. An ordinary bar’s low reaches it, the stop fills, and the position was risked on a distance the market covers routinely. In the narrowest fifth, the band was reached within five sessions on about two days in five.
The second failure is the reverse: a stop at the band after a volatile stretch. The band can sit five ATRs away, so the loss when it is hit is far larger than planned.
A third is comparing raw readings between instruments priced differently. Both are in price units.
A fourth is reading the squeeze as directional. It says closes bunched, not which way they will leave.
And a fifth is leaving both lookbacks at defaults without checking them. They were set elsewhere.
Related
Bollinger Bands covers the close-based envelope. Average true range covers the whole-bar measure. And standard deviation covers the statistic the bands are built from.
A stop has to survive the bar, gap included, and that is what average true range reports. Measure your stop in ATRs even if you draw Bollinger Bands, so that the distance you risk does not change every time the closes bunch up or spread out.
— Michael Whitman
This page is educational, not financial advice. Test every idea on your own charts before risking money.