Bollinger Bands: A Volatility Measure Dressed as a Price Channel

A Moving Average With Statistical Edges

Fidelity credits John Bollinger with developing the indicator and describes it as a type of price envelope, plotting bands at standard deviation levels above and below a simple moving average.

Standard settings are a 20-period average with bands two standard deviations away, though these are adjustable. Those are Bollinger's original parameters and remain the reference.

Why the bands breathe

Standard deviation measures dispersion in the recent price series. When a stock has been moving in a narrow range, dispersion is low and the bands contract toward the average. When it has been swinging, dispersion rises and the bands expand.

The width therefore carries information the middle line does not. Narrow bands say the recent past has been quiet; wide bands say it has not. Neither says anything about direction.

Touching a Band Is Not a Signal

This is the misreading the indicator invites and its documentation explicitly corrects.

Fidelity states plainly that the pair of bands is not intended to be used on its own, and should be used to confirm signals given with other indicators. It adds that strong trends can push prices outside the bands for extended periods, which may warrant additional research before taking profits.

Read those together and the popular interpretation collapses. If price can sit outside the upper band through an entire advance, then a band touch is a description of a strong move rather than a warning about one.

The reversal that negates itself

Fidelity adds a further caution: if prices immediately move back inside the band after escaping it, the suggested strength is negated. And traders should watch for false moves in the opposite direction that may reverse before a proper trend begins.

Both cautions describe the same underlying problem. The band tells you a move is large relative to recent volatility. Whether it continues or reverses is exactly the question the bands cannot answer, and the two outcomes look identical at the moment of the touch.

What Two Standard Deviations Actually Assumes

The choice of two standard deviations is where a statistical intuition can mislead.

In a normal distribution, roughly 95% of observations fall within two standard deviations of the mean, which is where the sense that price "should" stay inside the bands comes from. Financial price changes are not normally distributed. Large moves occur far more often than a normal distribution predicts, which is precisely why extended excursions outside the bands are ordinary rather than exceptional.

The standard deviation here is also calculated from the same 20-period window as the average, so it is a measure of recent behavior, not of the security's long-run character. A stock emerging from a quiet stretch will have tight bands that a genuinely ordinary move can breach.

Width Carries More Information Than Position

Because the bands respond to dispersion, their separation is itself a reading.

A sustained contraction says the security has been trading in an unusually narrow range for its recent history. That condition tends not to persist indefinitely — volatility clusters and alternates — but the contraction says nothing about which direction the eventual expansion will take.

This is the honest version of the popular idea that a squeeze precedes a breakout. Something usually follows a period of unusual quiet. The bands offer no information about what.

What the Three Lines Say Together

Reading the components as a set rather than watching for touches produces a more defensible picture.

The middle band is doing more work than it appears

The 20-period average is not decoration. It is the reference every other reading is measured against, and its slope carries the only directional information in the construction.

A rising middle band with price persistently in the upper half describes something different from a flat middle band with price oscillating across it, even when both produce band touches. The touches look similar; the underlying conditions are not.

The Structural Blind Spots

• No direction. Every reading concerns dispersion. Up and down moves of equal size affect the bands identically.
• No volume. A move on heavy participation and one on almost none produce the same band behavior.
• Window dependence. The 20-period default sets what counts as normal; a different window redefines every reading.
• Thin securities distort it. Where liquidity is low and spreads wide, a few trades can widen the bands without reflecting genuine volatility.
• Market structure bounds the input. Volatility bands set by market rules cap what prices can print, so the series the indicator measures is already constrained by rules.

The Cost of Trading Every Touch

An indicator producing frequent signals faces a cost hurdle before it produces anything else.

The regulator treats this as serious enough to mandate disclosure. FINRA Rule 2270 requires a standardized risk statement before an active-trading account opens, furnished individually to non-institutional customers and posted conspicuously on the firm's website.

Two elements of that statement bear on band trading specifically. Customers must be cautioned about material promoting outsized gains, which describes a considerable share of indicator marketing. And the rule works through an example in which commissions alone require $111,360 of annual profit to reach break-even, before any spread is counted.

Spreads are the larger cost for most retail traders now that commissions have compressed, and they are charged on both sides of every round trip regardless of whether the band touch was followed by anything.

Band touches are common. In a volatile stock they occur repeatedly within a month, and each acted-upon touch costs the spread twice.

Why the Statistical Framing Invites Overconfidence

The deeper issue with this indicator is presentational. Because it is built from a standard deviation, it carries an air of statistical rigor that its application does not support.

A standard deviation computed over twenty observations is a small-sample estimate of a quantity that is itself changing. It describes the dispersion of the last twenty periods, and it is used to make claims about the next one. Nothing in the construction justifies that step.

This matters because the framing makes band touches feel like measured events rather than descriptive ones. A price two standard deviations above a twenty-period mean sounds like a statistically significant departure. It is a description of the last twenty closes, calculated in a way that assumes a distribution the data does not follow.

Using Them as Intended

Bollinger's construction, and the guidance published alongside it, both point to the same use: context rather than instruction.

The bands answer whether the current move is large relative to the security's recent behavior, and whether that behavior has been unusually calm or unusually turbulent. Both are real observations that a raw price chart conveys poorly.

A reasonable working use is as a filter rather than a trigger. If the bands are unusually narrow, conditions are quiet and any breakout signal from another indicator is arriving into a market that has not been moving. If they are unusually wide, the security is already volatile and position sizing matters more than entry precision. Neither reading tells you what to do; both change how much confidence a separate signal deserves.

They do not answer whether to buy or sell, and Fidelity's instruction to use them for confirmation alongside other indicators is the operative guidance rather than a disclaimer. The same publisher gives comparable advice for its other indicators, which is itself informative: the tools are designed to be read together.

One further practical note. Because band width tracks recent volatility, the indicator recalibrates itself after every regime change. A stock that becomes permanently more volatile will, within twenty periods, have wider bands that treat the new behavior as normal.

That adaptiveness is a strength for describing current conditions and a weakness for detecting change, since the measure quietly absorbs exactly the shift a reader might most want flagged.

The Bottom Line

Bollinger Bands plot a 20-period average with edges two standard deviations away, and those edges move with recent volatility rather than marking levels price should respect.

The single most important correction to the popular reading comes from the documentation itself: strong trends push price outside the bands for extended periods, and the bands are not meant to be used alone. A touch of the upper band during a powerful advance is the indicator working correctly and reporting a large move, not warning of a reversal.

Read width as a volatility statement, treat position as context, require confirmation from something that measures a different thing, and check the transaction costs before acting on any of it.

For a bounded momentum measure, see RSI; for a trend construction with a different failure mode, see MACD; for what each trade costs, see the spread.