Bollinger Bands are the most commonly misread indicator on this list, because they look like boundaries. They are not boundaries. They are a measure of how unusual the current price is relative to its own recent behaviour — and "unusual" happens constantly.
The construction
Three lines, from standard settings of 20 and 2:
- Middle band — a 20-period simple moving average.
- Upper band — the middle band plus two standard deviations of the same 20 closes.
- Lower band — the middle band minus two standard deviations.
Standard deviation is a measure of dispersion. When recent closes have been tightly clustered it is small and the bands sit close together; when they have been scattered it is large and the bands flare apart. The bands are therefore a volatility envelope that adapts automatically — that adaptation is the whole idea.
The "95%" claim is wrong
You will read that two standard deviations contain about 95% of observations. That is true of a normal distribution. Financial returns are not normally distributed — they have fat tails, meaning extreme moves happen far more often than the bell curve predicts.
So price sits outside the bands more often than the statistic implies, and it does so in clusters, because volatility is persistent: a violent day is usually followed by another. Any strategy built on "price rarely leaves the bands, so fade it" rests on a distributional assumption the market does not honour.
Touching a band is not a signal
This is the central point. In a strong trend, price does not just touch the upper band — it walks it, riding along the outside for days or weeks. Each touch is the indicator correctly reporting a powerful move, and each fade of that touch is a loss.
A band touch means price is statistically extended relative to the last 20 periods, and extended is what a trend looks like from the inside. The touch is a description, not an instruction — a distinction that matters more here than anywhere else, because the visual of price hitting a line is so suggestive.
Mean reversion at the bands only has a foundation in a range, where a mean exists to revert to. Establish the regime first, then read the bands within it.
The squeeze — the useful part
When the bands contract sharply, recent closes have clustered tightly: volatility has collapsed.
This matters because volatility is cyclical in a way direction is not. Quiet periods reliably give way to active ones and active periods to quiet ones. A squeeze therefore carries a genuine probabilistic statement — a large move is more likely than usual — and it is one of the few things on a chart that does.
What the squeeze does not tell you is direction. It is common to see a squeeze break one way, trap everyone who chased it, and reverse hard the other. Traders who treat the first candle out of a squeeze as the answer get caught by this repeatedly. Waiting for the break to hold, or structuring the entry so both directions are acceptable, is the sane response.
%B and bandwidth
Two derived readings are worth knowing:
- %B expresses where price sits within the bands on a 0–1 scale (0 at the lower band, 1 at the upper, above 1 outside). It makes band position comparable across instruments.
- Bandwidth is the width of the bands relative to the middle line. Plotted over time it shows the volatility cycle directly — and a bandwidth reading at multi-month lows is the squeeze, stated numerically rather than judged by eye.
ATR — the other volatility tool
Average True Range measures the typical size of a period's full range, including gaps. It says nothing about direction and nothing about position — only about magnitude.
For the job most traders actually need volatility for, ATR is the better tool. Stop distances should scale with how much the instrument moves: a 20-pip stop is loose on a quiet day and absurdly tight on a volatile one. Sizing stops and targets in multiples of ATR keeps risk consistent as conditions change, which fixed pip distances cannot do.
A reasonable division of labour: Bollinger Bands to read the volatility regime, ATR to size the trade within it.
What to take from this
- Bands are a moving average ± 2 standard deviations — a volatility envelope, not a boundary.
- The 95% statistic assumes a normal distribution that markets do not follow.
- Price walks the band in trends; fading touches only has a basis in a range.
- The squeeze predicts a large move, never its direction.
- Use ATR, not band width, to size stops — risk should scale with actual range.
The last lesson of this module covers Fibonacci retracements — the most popular tool in the toolbox and the one with the weakest theoretical foundation.