Bollinger Bands vs Standard Deviation
Bollinger Bands take a statistical measure of how spread out recent closes have been and plot it a chosen number of deviations either side of a moving average. Standard deviation is that underlying measure, so the bands are simply the number drawn on a chart.
These are not two tools. One is a statistical measure and the other is that measure drawn around a moving average, which makes the comparison a question about presentation.
What each one is
Standard deviation measures how spread out recent values have been. Applied to closing prices it reports dispersion in price units. Standard deviation covers it.
Bollinger Bands plot that measure on the chart. A moving average with lines a chosen number of deviations either side. Bollinger Bands covers the construction.
So one contains the other. Everything the bands do is the underlying measure made visible, and every limitation of the measure is inherited whole.
Where they differ
Whether it is a picture or a number. The bands are readable at a glance; the raw measure is a value you compare against its own history.
What is added by plotting. A centre line and a multiplier. Those are choices layered on top of the measure, and both change what a band touch means.
How each is used. The bands for touches and squeezes; the raw number usually inside position sizing or a filter rather than as a chart feature.
How comparable across instruments. The raw measure in price units is not comparable between markets; expressed relative to price it becomes so, which the bands do not do for you.
Where they agree
They are the same calculation. Nothing in the bands originates outside the measure, so the two cannot disagree in any meaningful sense.
Both are dominated by outliers. One large bar moves a dispersion figure sharply, which is why the envelope expands immediately after the move that would have hurt you.
Both read closes only. Gaps and wicks do not enter the calculation at all — on this site’s shared series the largest single bar range was 2.338 against a median of 0.493, and neither sees that directly.
And both lag. Every value comes from bars that have already printed, and neither anticipates anything.
Which one to use
Use the bands when you want the measure on the chart. Seeing dispersion in place, relative to price, is genuinely easier to read than a number in a panel.
Use the raw measure when a rule needs a number. Position sizing, a volatility filter or a comparison across instruments all want the value rather than a picture of it.
Use the raw measure when you want to normalise. Dividing by price makes it comparable between markets, which the bands do not do and cannot show.
And when somebody presents the two as alternatives, they have missed the relationship. One is drawn from the other, and any disagreement between them is an error somewhere.
Why the outlier sensitivity matters
Because dispersion is defined by the extremes. A single violent bar moves the measure a great deal, so both the number and the envelope expand right after the move rather than before it.
And because closes hide the rest of the bar. A day that swung widely and closed flat barely registers, which is exactly the situation average true range was designed to capture.
What a band touch actually means
That price is a stated number of deviations from its own average. That is the entire claim, and the multiplier decides how often it happens.
Not that price is extreme. The bands are calibrated so touches are frequent, and in a trend price can ride the edge for many bars.
Not that a reversal is due. On this site’s shared series direction runs average 2.01 bars with a longest of 11, and nothing about a touch shortens or lengthens that.
And nothing about where a stop belongs. The ninetieth percentile bar range here is 1.101, so a stop at a band edge is a distance chosen by a formula rather than by structure.
The original data
Of the 24,971 unique videos in research/search-study-corpus.jsonl, no title compares these two
directly, and only 1 names standard deviation at all — this pair is constructed from subjects the
corpus covers very unevenly. Separately, Bollinger Bands appear in 259 titles at a median of 5,178
across 197 channels. The counts come from site/corpus_count.py.
259 videos on the picture and 1 on the measure underneath it. The most widely taught volatility tool in the corpus rests on a statistic that essentially nobody explains, which is a striking gap given how much of the tool’s behaviour follows directly from it.
The answer to the question on that chart is that it is a stated distance from an average. The bands were calibrated so this happens often — so the touch describes where price sits and predicts nothing.
When it fails
The failure is treating a band touch as an extreme, and a trend punishes it every bar. Price reaches the upper edge, which reads as stretched, so a fade is taken. The bands are constructed so touches are common, and in a strong move price rides the edge while the envelope widens beneath it. Each new bar looks more extreme than the last, which makes the position feel more justified as it loses.
The second failure is comparing raw values across instruments. They are in price units.
A third is using a close-based measure on a gappy market. It cannot see gaps.
A fourth is placing a stop at a band edge. The largest bar range here was 2.338.
A fifth is treating the squeeze as directional. It says nothing about which way.
And a sixth is running the bands and the raw measure together. That is one calculation twice.
Related
Bollinger Bands covers the plotted version. Standard deviation covers the measure underneath. And average true range covers the alternative that reads the whole bar.
Knowing that the bands are just the measure drawn on the chart changes how you read them. A touch is not a signal — it is price being a stated distance from its average, which the calculation was designed to make a common occurrence.
— Michael Whitman
This page is educational, not financial advice. Test every idea on your own charts before risking money.