Skip to content

Resource · Public

Bollinger Bands: Volatility, Band Width, and False Signals

Learn to separate Bollinger Band width from direction, calculate an example, and record touches, squeezes, and failed breaks fairly.

Published

Volatility bands widen and narrow around candles, with calipers illustrating changing band width.
Conceptual illustration: volatility bands widen and narrow around candles, with calipers illustrating changing band width.
Image credit

Conceptual illustration: volatility bands widen and narrow around candles, with calipers illustrating changing band width.

FavAcademy · Original commissioned AI illustration; public use authorized

Public summary

Bollinger Bands place a moving envelope around recent prices. A touch can occur during a reversal or a continuing trend. This lesson separates width from direction, works through a transparent numerical example, and builds a log that includes failed breaks and repeated band touches.

Ask three separate questions

When an envelope surrounds price, it is easy to turn the whole picture into one judgment. Instead, ask where the middle line is, how far apart the outer lines are, and where the latest price sits relative to them. These are different observations. A rising center line can coexist with narrowing bands. A falling price can coexist with expanding bands. Bollinger Bands commonly use a moving average and a standard-deviation distance above and below it. Record the actual period and multiplier rather than assuming every platform uses your preferred settings. Our shorter illustrative window below exists to reveal the mechanism, not to recommend a configuration.

Calculate a tiny envelope

Take five invented closes: 98, 99, 100, 101, and 102. Their mean is 100. The squared deviations from the mean are 4, 1, 0, 1, and 4, totaling 10. Using the population standard deviation for this example gives the square root of 10 / 5, or approximately 1.4142. With a multiplier of two, the upper band is about 102.8284 and the lower band about 97.1716. Their distance is about 5.6569 price units. A charting package using a different deviation convention can differ; document that convention before comparing values. Neither the band boundaries nor the mean are independent observations of future prices.

Turn width into a comparable description

For a positive center line, one common normalized width calculation is (upper band − lower band) / middle band. In our example that is approximately 0.056569, or 5.6569% when expressed as a percentage. This lets you distinguish an absolute distance from a distance relative to the instrument's price scale. It does not make every instrument or interval directly comparable. A width measured from hourly closes and one measured from daily closes answer different questions. Compare like with like, inspect the historical distribution under fixed settings, and say whether your software displays a ratio or a percentage. A factor-of-one-hundred display mistake can look like a dramatic volatility change.

Follow a price that keeps touching

Imagine an invented sequence where price repeatedly closes near a rising upper band while the center line also rises. A learner who sells every touch is repeatedly opposing that observed move. Another learner might view a band touch as a candidate for trend continuation. The envelope alone cannot settle the argument because both interpretations require additional rules and evidence. Write the exact proposed action, its invalidation, and its cost assumptions before reviewing subsequent bars. Also distinguish an intrabar excursion from a completed close. The picture can change while the bar is still forming, and a historical completed chart can hide those temporary appearances.

Treat contraction as a condition, not a direction

Narrower bands describe reduced dispersion within the calculation window. They do not tell you which side will break next, how far price will travel, or whether a break will persist. Make a two-sided exercise: mark a narrow interval, conceal the later data, and write a plan for an upward break, a downward break, and continued inactivity. Reveal one bar at a time. If price breaks upward and then returns inside, retain the original annotation and log the failure. Do not redraw the starting interval after seeing the result. This preserves the difference between information available at the decision and information supplied by hindsight.

Do not turn standard deviations into odds

A band based on two standard deviations is not automatically a 95% forecast interval for the next close. That shortcut requires assumptions about a distribution and its stability that this rolling descriptive calculation does not establish. Prices can cluster near an edge, move outside it repeatedly, and alter the bands as new observations arrive. A particularly large close affects both price location and the subsequent envelope. If you want to estimate the frequency of a defined event, count it in a suitable dataset with fixed rules. State the sample, dependence concerns, and uncertainty rather than attaching a familiar probability to a chart decoration.

Keep a band-event record

Create columns for time, settings, center, upper, lower, width convention, touch type, contextual rule, next eligible action, and later outcome. Include events that led to no trade as well as those that met your rule. Keep risk limits and transaction costs separate from the indicator description. For a useful practice session, compare repeated touches during one directional sequence with a false break during a sideways sequence; both can be drawn with invented data. Explain what each event reveals about the past window and what remains unknown. The objective is to read a changing envelope accurately before asking whether a complete strategy built around it deserves further testing.

Risk disclosure

Examples use invented prices and a stated deviation convention. A band touch or squeeze does not establish direction or future odds.