How the bands are built
Only three components, all derived from a moving average:
Middle = SMA(20) | Upper = middle + (2 × standard deviation) | Lower = middle − (2 × standard deviation)
Standard deviation is a statistical measure of how dispersed prices are around their average. When prices swing violently the deviation grows and the bands widen automatically; when things calm down they contract. This means the bands breathe with the market rather than being fixed lines — the core of John Bollinger's idea: a channel that adapts itself to conditions.
Because the middle is a moving average, everything true of the moving average regarding lag is true of Bollinger too. And its volatility basis makes it closely related to ATR: both measure volatility, but ATR gives it an absolute number while Bollinger draws it around price.
Why touching a band is not a signal
This is the most repeated error about the indicator: "price touched the upper band, so it is overbought — sell." In reality the upper band is neither a ceiling nor resistance; it is simply a statistical description of how far price has moved from its average. And in a strong uptrend, being far from the average is the expected normal behaviour, not an anomaly.
The phenomenon is known as "walking the band": in a strong trend price hugs the upper band candle after candle as it climbs. Anyone selling the first touch has opened a position against the strongest move in the market, then compounds the error by selling again on the second touch.
| Context | What an upper-band touch means | The logical response |
|---|---|---|
| Sideways range, narrow bands | A temporary excursion likely to revert to the middle | A reasonable mean-reversion case, with confirmation |
| Uptrend, wide bands | Trend strength, not exhaustion | Selling here fights the trend |
| First touch after a breakout | Usually the start of a move | Wait rather than fade |
In short: a band touch is a question, not an answer. The answer comes from context — is the market trending or ranging? This is why Bollinger is always read alongside price structure and support and resistance, never alone.
The squeeze: what it actually predicts
When the bands contract to their narrowest in some time, this is called a squeeze and means volatility has reached an unusually low level. Since volatility tends to alternate between quiet and active phases, a squeeze often precedes a wide move.
But note the decisive detail most explanations skip: a squeeze predicts the size of the move, not its direction. It says "a big move is coming", not "up" or "down". Buying merely because you see a squeeze is betting on a direction the indicator never provided. The correct response is preparation rather than prediction: mark the breakout levels above and below the range and act with whichever side actually resolves.
Two practical measures beat eyeballing here: Bandwidth measures the distance between the bands relative to the middle, quantifying the squeeze, and %B locates price within the bands (0 at the lower, 1 at the upper, 0.5 at the middle). Using %B is far more precise than saying "price is near the band" because it gives a number comparable across instruments and time.
When Bollinger does not apply: limits and mistakes
The statistical limit first, which almost nobody states: ±2 standard deviations is said to contain about 95% of movement, but that holds only if market returns were normally distributed — and they are not. Traded markets have "fat tails": extreme moves happen far more often than a normal distribution assumes. So do not treat price leaving the bands as a vanishingly rare event; in crises and major news it happens regularly.
- Do not use it as a standalone signal — Bollinger describes a state, it does not make a decision.
- Do not assume mean reversion in a strong trend; walking the band can last longer than your account can withstand.
- Beware narrow bands on small timeframes, where a squeeze becomes an almost permanent condition with no meaning.
- Changing settings changes the meaning: (20, 2) is the default; 1 deviation makes touches near-constant, 3 makes them very rare.
- It misreads gaps: a large gap distorts the standard deviation for subsequent periods.