Bollinger Bands are a volatility indicator plotted around price. The standard version uses a 20-period simple moving average as the middle band. The upper and lower bands sit two standard deviations above and below it.
The bands answer two useful questions:
- Is the current price relatively high or low compared with its recent range?
- Is recent volatility contracting or expanding?
They do not tell you that price must reverse at a band. They also do not predict the direction of the next breakout.
This is how I read the basic picture:
| What the bands show | What it means |
|---|---|
| Price touches the upper band | Price is high relative to the previous 20 closes. It is not automatically a sell signal. |
| Price touches the lower band | Price is low relative to the previous 20 closes. It is not automatically a buy signal. |
| Bands narrow | Recent volatility has fallen. Direction is still unknown. |
| Bands widen | Recent volatility is increasing. |
| Price repeatedly follows one band | A trend may be strong enough to keep price near that side of the envelope. |
What the three Bollinger Bands measure
Middle band
The middle line is normally a 20-period simple moving average. It describes the average closing price over the selected window.
It is not a separate prediction line. If price rises, the average eventually rises with it. The same lag applies when price falls.
Upper band
The upper band is the middle band plus two standard deviations. It moves farther from the average when recent closes become more dispersed.
Lower band
The lower band is the middle band minus two standard deviations. Upper and lower bands use the same calculation, so they normally expand and contract together.
Middle band = 20-period SMA
Upper band = Middle band + (2 x 20-period standard deviation)
Lower band = Middle band - (2 x 20-period standard deviation)The default settings are written as 20, 2: 20 periods and two standard deviations.
An upper-band touch is not a sell signal
The upper band shows that price is high relative to its recent average. It does not say that price is about to fall.
AAPL closed above its upper Bollinger Band on November 17, 2021, at 153.49. Four more closes followed above the band. Twenty trading sessions after the first one, AAPL closed at 172.26, about 12.2% higher.
Selling only because price reached the upper band would have meant fighting the move. John Bollinger's own rules make the same distinction: a band tag is a tag, not a signal, and price can walk along a band during a trend.
When I see repeated upper-band tags, I check:
- Is price still making higher highs and higher lows?
- Is the middle band rising?
- Are pullbacks holding near or above the middle band?
- Is BandWidth expanding or at least staying elevated?
Before I treat the move as bearish, I want to see a failed breakout or a break of the latest higher low. The band touch alone is not enough.
The same logic applies in a downtrend. Price can keep following the lower band while the middle band slopes down.
What a Bollinger Band squeeze tells you
A squeeze happens when the bands contract. It shows that recent price variation is low relative to the same market's recent history.
On April 21, 2022, the 20-period BandWidth for GBP/USD had fallen to 1.46%. The bands were close together. By May 12, BandWidth had reached 9.06%, roughly 6.2 times wider.
GBP/USD moved lower in this example. It closed at 1.3026 on April 21 and at 1.2462 on May 19, a decline of about 4.3%.
The squeeze identified compression. It did not predict a downside breakout. Direction only became visible after price left the range and started forming lower highs.
I do not use a universal BandWidth threshold. A 3% reading may be quiet for one asset and normal for another. I compare it with the same instrument and timeframe, usually over a fixed historical window.
How to use Bollinger Bands in trends and ranges
The same band touch can mean different things because the surrounding price structure is different.
In a trend
Price may spend more time between the middle and upper band during an uptrend, or between the middle and lower band during a downtrend.
In this environment, I use the bands to read trend position and volatility. A return toward the middle band may be a normal pullback. A close outside a band may show expanding momentum rather than exhaustion.
In a range
Band touches become more useful for mean-reversion ideas when price has already established horizontal boundaries.
In a range, I want three things to line up: an established horizontal boundary, a move through the band, and a close back inside both the band and the range. The rejection matters more than the touch. Its high or low also gives me a clear invalidation level.
If price holds outside the range while the bands expand, the mean-reversion idea is no longer valid. I do not keep fading the move because price remains near an outer band.
Bollinger BandWidth and %B
The three lines are the visible part of the indicator. Two derived values make the same information easier to compare.
BandWidth
BandWidth measures the distance between the upper and lower bands relative to the middle band.
BandWidth = (Upper band - Lower band) / Middle bandA falling value means the bands are contracting. A rising value means they are expanding. The lower panel in both charts above plots BandWidth as a percentage.
%B
%B describes where price sits inside or outside the bands.
%B = (Price - Lower band) / (Upper band - Lower band)- 1.0: price is at the upper band.
- 0.5: price is at the middle band.
- 0.0: price is at the lower band.
- Above 1.0: price is above the upper band.
- Below 0.0: price is below the lower band.
%B does not create a new signal. It converts the price's location into a number that can be logged and tested.
Bollinger Bands settings
| Setting | Default | What it changes |
|---|---|---|
| Period | 20 | The number of candles used for the SMA and standard deviation. |
| Deviation | 2 | The distance of each outer band from the middle average. |
| Price source | Close | The data used in the calculation. |
Shortening the period makes all three bands react faster. It also makes them more sensitive to ordinary price noise. Increasing the deviation moves the outer bands farther away and reduces the number of outside closes.
The period counts candles, not days. A 20-period setting covers 20 daily closes on a daily chart and 20 five-minute closes on a five-minute chart.
I start with 20 and 2 because they are the shared reference. I change them only when the complete trading rule can be tested on the intended instrument and timeframe. Adjusting the bands until they fit one historical chart is not a test.
My Bollinger Bands checklist
- Market structure: trend, range or compression?
- Band event: touch, outside close, squeeze or expansion?
- Middle band: rising, falling or flat?
- BandWidth: low, rising or already elevated relative to this market?
- Price evidence: breakout, rejection or no confirmation?
- Invalidation: which price level proves the idea wrong?
- Costs: is the expected move large enough after spread, fees and slippage?
I do not use the upper and lower bands as automatic buy and sell buttons. Their value is in showing where price sits relative to its recent average and how the market's volatility is changing.
