Bollinger Bands: Squeezes, Breakouts and Head Fakes

John Bollinger designed his bands in the early 1980s to answer a question moving averages cannot: is this market quiet or loud right now? The bands widen when prices swing and tighten when they settle. Most people use them to call tops and bottoms at the outer lines. The more interesting information is in how far apart the lines are.

How the bands are drawn

1. Middle band: the 20-day simple moving average of closes.
2. Standard deviation: how far those 20 closes typically sit from their average.
3. Upper and lower bands: the middle band plus and minus two standard deviations.
4. Bandwidth: (upper − lower) ÷ middle, shown as a percentage. This is the squeeze gauge.

Here is AMD on September 30, 2025, after a month of going nowhere. Its 20 closes ranged from 151.14 to 162.32.

On its own 7.87% means little; bandwidth only makes sense compared with the same stock's history. AMD's bandwidth had not been that narrow since July 1, 2024, 313 sessions earlier. In early September the same measure had been above 20%. That contrast, not the absolute number, is what traders mean by a squeeze.

Case 1: AMD, from squeeze to expansion

The resolution came in three steps, and only the last one was obvious.

• October 2: AMD closed at 169.73, above the upper band at 167.32. Volume was 1.25 times its 20-day average, nothing dramatic.
• October 3: it slipped back inside the bands, closing at 164.67. A trader watching only for band breaks could easily have called this a failure.
• October 6: after AMD announced a multi-year chip supply agreement with OpenAI, the stock opened at 226.45, 37.5% above the October 3 close, and finished at 203.71 on volume 5.7 times its average.

Bandwidth went from 7.87% to 24.35% in one day and kept expanding to 66.44% by October 16. The squeeze did its job: it flagged that AMD was unusually coiled. It did not tell anyone about the news, the size of the gap, or the date. Nobody could have traded the October 6 open from the September 30 reading, and it would be dishonest to present the squeeze as if it predicted that jump.

Walking the band

After the gap AMD closed above its upper band four sessions in a row, from October 6 to October 9. A close outside the upper band is often read as "stretched, sell". In a genuine expansion it is the opposite: price riding the outer band is what a strong trend looks like on this indicator. The first close back inside, on October 10 at 214.90, was the first hint the burst was cooling.

Case 2: Apple's head fake, September 2024

Bollinger himself warns about "head fakes": after a squeeze, price often pokes out one side first, then reverses and makes the real move the other way. Apple gave a tidy example.

A trader who shorted the break below the lower band on September 16 would have been 5.8% offside within three sessions. The lesson is not that downside breaks fail; plenty follow through. It is that the first move out of a squeeze deserves less trust than the second, and that a stop placed just inside the band keeps a head fake from becoming a large loss.

The 95% myth

You will often read that about 95% of prices stay inside two-standard-deviation bands. That figure comes from the normal distribution, and daily stock prices are not normally distributed. We checked closes from 2023 through 2025:

Roughly one close in eight lands outside the bands, even for a slow stock like Coca-Cola. A close beyond a band is less rare than the textbook suggests, which is one more reason not to treat every touch as an extreme.

Reading the bands, in order of usefulness

• Bandwidth relative to the stock's own past: the lowest reading in six months or more marks a squeeze worth watching.
• Which side price leaves from, and whether the break holds for more than a session. Expect a head fake as a real possibility.
• Whether volume confirms the break. AMD's real move came with 5.7 times normal volume; the October 2 poke came with 1.25 times.
• In a range, a tag of the outer band near support or resistance can mark a turn. In a trend, band tags are normal.

Bands pair well with a momentum reading such as RSI: the bands tell you how wide the swings are, RSI tells you how one-sided the recent closes were. Next time you see bandwidth at a multi-month low, guess the direction before the break and write it down. Checking those guesses later is humbling, and it is the fastest way to stop treating a squeeze as a directional signal.