Bollinger Bands Explained: How to Read Them, and What Our Test Found

Bollinger Bands are three lines drawn around price: a moving average in the middle and two bands a fixed number of standard deviations above and below it. They do not predict direction. They answer a narrower question that most traders never ask out loud: is the current price unusual compared with the last few weeks, and is the market getting quieter or louder? Read that way, they are one of the most useful tools on a chart. Read as "upper band means sell, lower band means buy", they are one of the most expensive.

The three lines, and what each one measures

The standard setting is 20 periods and 2 standard deviations:

• Middle band: the 20-period simple moving average of the close.
• Upper band: middle band plus 2 times the standard deviation of the last 20 closes.
• Lower band: middle band minus 2 times that standard deviation.

Standard deviation is a measure of how far the closes have been spreading around their average. When price moves calmly, it is small and the bands pull in. When price swings hard, it grows and the bands push out. So the distance between the bands is a live reading of volatility, and the position of price inside them is a reading of how stretched it is relative to that volatility.

A worked example

Say Bitcoin closes at 64,000 on the daily chart. The 20-day average is 62,000 and the standard deviation of the last 20 closes is 1,500.

• Upper band: 62,000 + 2 × 1,500 = 65,000
• Lower band: 62,000 − 2 × 1,500 = 59,000

Two derived numbers make the picture precise.

%b tells you where price sits inside the bands, from 0 at the lower band to 1 at the upper band:

%b = (close − lower) / (upper − lower) = (64,000 − 59,000) / 6,000 = 0.83

Above 1 means price closed outside the upper band; below 0 means outside the lower band; 0.5 is the middle line.

Bandwidth tells you how wide the bands are relative to price:

bandwidth = (upper − lower) / middle = 6,000 / 62,000 = 9.7%

Bandwidth on its own means little. Compared with its own history it means a lot: if the lowest reading of the last six months is 4%, a market at 4.2% is unusually quiet, and quiet markets do not stay quiet. That is the idea behind the [Bollinger squeeze](/blog/bollinger-squeeze).

The mistake almost everyone makes: treating a touch as a signal

The most common Bollinger "strategy" is to sell at the upper band and buy at the lower band. It works in a range and fails badly in a trend, for a simple reason: in a strong trend, price walks along the band. Each new high pushes the average up and widens the bands, so price can tag the upper band ten times in a row while rising the whole way.

John Bollinger, who designed the tool, says the same thing in different words: a tag of the band is not a signal by itself. It is information about where price is relative to recent volatility. What you do with it depends on whether the market is trending or ranging, and the bands alone do not tell you which.

A practical way to separate the two:

1. Ranging: the middle band is roughly flat, bandwidth is stable, and price keeps returning to the middle line after touching either band. Fading the bands has a chance here, with a stop beyond the band.
2. Trending: the middle band slopes clearly, price stays between the middle and outer band on one side, and pullbacks stop at the middle line. Here the upper band is evidence of strength, not of exhaustion. The pullback to the middle band is usually the more interesting moment, not the touch of the outer one.

If you cannot tell which of the two you are looking at, that is an answer too: the [trend or range](/learn/technical/trend-or-range) question comes before any band rule.

The textbook "95% inside" figure does not hold

You will often read that about 95% of prices fall inside the bands. That figure assumes returns follow a normal distribution. Market returns do not: they have fat tails, which means large moves happen far more often than a bell curve allows. Expect noticeably more closes outside the bands than 5%, especially in crypto and around news. Treat a close outside the band as "unusual", not "nearly impossible".

Settings: when to change them, and how

The 20/2 default is a sensible starting point for daily charts. If you change the length, Bollinger's own guidance is to adjust the width with it so the bands contain a similar share of price:

| Length | Width |

|---|---|

| 10 | 1.9 standard deviations |

| 20 | 2.0 standard deviations |

| 50 | 2.1 standard deviations |

Two warnings. First, tuning the settings until past signals look good is curve fitting; the settings that looked best on last year's chart are rarely best on next year's. Second, a shorter length makes the bands react faster and produce more tags, which means more noise, not more information.

What our own test found

We ran a large test of textbook conditions on the 20 largest crypto assets, on daily, weekly and 4-hour charts, with trading costs doubled to stay conservative. The question was narrow: at the moment a trend-following entry happens, does a given condition being true change how that entry turns out? We tested 102 conditions on three timeframes, 306 tests in total, and corrected for the fact that testing that many things will produce lucky winners by chance.

Three Bollinger conditions were in the set:

• %b above 0.5 on the 4-hour chart (price in the upper half of the bands) was one of only six conditions out of 306 that showed a statistically reliable difference. Entries taken while it was true did better on average than entries taken while it was false.
• Close above the upper band did not pass.
• A squeeze release (bands widening after a squeeze, with price above the middle) did not pass.

Then the honest part. When we used the surviving condition as a filter, meaning we skipped every entry where it was false, the result was worse than not filtering in most of the test periods. The condition was real information, but blocking trades with it cost more good trades than it removed bad ones.

The lesson generalises well beyond Bollinger Bands: a condition that describes good trades is not automatically a rule that produces them. "Price is in the upper half of its bands" is a useful description of short-term strength. It is not a strategy.

How to actually use Bollinger Bands

• As a volatility gauge. Bandwidth tells you whether to expect calm or movement, which matters for [where you put a stop](/learn/risk/where-to-put-a-stop) and how large the position can be.
• As a context check. %b tells you whether you are buying something already stretched. Buying at %b of 1.2 means your stop has to be far away or very tight, and both have costs.
• Together with a different kind of indicator. Bands are built from price, so pairing them with another price-based oscillator mostly repeats the same information. A volume-based measure, or a trend filter on a higher timeframe, adds something new. Piling on more lines that all say the same thing is [indicator overload](/learn/technical/indicator-overload).

Frequently asked questions

What are the best Bollinger Band settings? There are no best settings, only consistent ones. 20 periods and 2 standard deviations is the standard and works as a reference on most timeframes. If you change the length, change the width with it (10 with 1.9, 50 with 2.1), and test the change on data you did not use to choose it.

Is touching the upper Bollinger Band a sell signal? No. In a range it can mark a stretched price; in an uptrend price can ride the upper band for weeks. A touch tells you price is high relative to recent volatility, nothing more.

What is the difference between Bollinger Bands and Keltner Channels? Both draw an envelope around a moving average. Bollinger Bands use standard deviation of closes, Keltner Channels use average true range. Bollinger Bands react more sharply to sudden moves; Keltner Channels are smoother. The squeeze setups that compare the two exist because the difference between them highlights unusually quiet periods.

Do Bollinger Bands work in crypto? They measure the same thing in any market. Crypto has fatter tails and more closes outside the bands, so band tags are less meaningful and volatility readings are more useful.

Indikora is an AI-powered trading coach for crypto, forex, gold and indices. It reads your own trades, shows the context behind each entry, and explains its signals instead of just issuing them. Try it free: https://indikora.com

This article is for educational purposes only and is not financial advice.