Bollinger Bands

Bollinger Bands were created by John Bollinger in 1983 and have become the most widely used volatility indicator in retail trading. They are also the indicator most likely to be misapplied: traders sell at the upper band, buy at the lower band, and lose money in trending markets. Bollinger himself has warned against this approach for four decades.

The bands measure volatility, not direction. They tell you when the market is calm and when it is excited; they do not tell you whether the next move is up or down. Using them as a directional indicator inverts what they are designed to do.

What Bollinger Bands actually measure

Bollinger Bands wrap a moving average in a statistical envelope. The middle band is a 20-period simple moving average (SMA). The upper and lower bands are placed two standard deviations above and below the SMA, computed over the same 20-period window.

Standard deviation is a measure of dispersion: how spread out the recent closing prices have been around their mean. When volatility rises, standard deviation rises, and the bands widen. When volatility falls, standard deviation falls, and the bands contract.

Because the bands are 2 standard deviations wide, roughly 95 percent of price action will historically stay within them - if returns were normally distributed, which they are not. The actual containment rate is closer to 88-93 percent. That detail matters: closes outside the band are slightly more common in real markets than the normal-distribution math suggests.

The formula

Middle band = 20-period SMA(close)
Upper band  = Middle band + (2 × σ)
Lower band  = Middle band - (2 × σ)

where σ = standard deviation of the last 20 closes

Standard deviation is computed as:

σ = sqrt( Σ(close_i - mean)² / N )

where the sum runs over the last N closes (here N = 20) and the mean is the 20-period SMA. Take the square root of the average squared deviation from the mean.

You can verify your platform's Bollinger Bands in a spreadsheet by computing the 20-period SMA, then for each of the 20 closes subtracting the SMA, squaring the difference, averaging those squares, and taking the square root.

A worked example

Suppose the last 20 closes of an instrument average to 100.00 (the SMA). The closes deviated from that mean by varying amounts. Compute each deviation squared, average them, take the square root, and you get a standard deviation of - let us say - 1.50.

Middle band = 100.00
Upper band  = 100.00 + (2 × 1.50) = 103.00
Lower band  = 100.00 - (2 × 1.50) = 97.00

Now imagine volatility doubles over the next several bars. Each new close is 3 to 4 points away from the new mean, on average. The recomputed σ jumps to 3.00:

Middle band ≈ 100.00 (the mean is roughly unchanged)
Upper band  = 100.00 + (2 × 3.00) = 106.00
Lower band  = 100.00 - (2 × 3.00) = 94.00

The bands widened from a 6-point spread to a 12-point spread. Even though the middle didn't move much, the envelope around it doubled. That widening is the indicator telling you the market regime has changed - the same setup that looked tight five bars ago is now operating in a much noisier environment.

How traders actually use Bollinger Bands

Three setups generate edge that survives across thousands of trades. The rest is noise.

1. The squeeze

When the bands contract to their tightest reading in months, volatility has collapsed. Markets do not stay calm forever - a squeeze almost always resolves into a directional move. The bands themselves do not tell you which direction; combine them with structure (a higher-timeframe range, key support or resistance) for that.

The "Bollinger Band Width" indicator (BBW = (upper - lower) / middle) quantifies how tight the bands are. A 6-month low in BBW is a textbook setup for the next significant move.

The trade after a squeeze:

1. Wait for the breakout - a close outside the upper or lower band on rising volume
2. Enter on the close or on a pullback that holds the broken band
3. Stop loss inside the squeeze (often back at the middle band)
4. Target the height of the squeeze projected from the breakout

2. The band-ride (or "walking the band")

In strong trends, price closes outside the upper band repeatedly without reversing. This is the band-ride, and it is the inverse of what most traders expect. Price closing above the upper band is not a sell signal in a trend - it is a sign of strength.

The way to differentiate: check the middle band. If the middle band (20-SMA) is sloping clearly up or down and price is repeatedly closing outside the band in the direction of that slope, you are in a band-ride. Fading that is a losing trade.

Counter-intuitive but consistent: the more violent a band-ride looks, the more reliable it tends to be.

3. Mean reversion to the middle band

In range-bound markets (where the middle band is roughly flat), price oscillating between the upper and lower bands creates clean mean-reversion setups:

• Touch upper band in a range → short toward middle band
• Touch lower band in a range → long toward middle band

The requirement is that the middle band is flat. If it is sloping, the trend is intact and these setups are counter-trend entries that will lose.

The trap most retail traders fall into

The default Bollinger Bands lesson says: "Sell when price hits the upper band. Buy when price hits the lower band." This is the most expensive Bollinger Bands sentence in retail trading.

In a trending market, this strategy systematically fades the trend. The bands are designed to expand to accommodate strong moves, but they cannot expand fast enough to stay ahead of a parabolic trend. So price keeps closing outside the band, and the trader keeps adding to a losing counter-trend position.

The fix is to read the middle band first:

• Middle band rising clearly: bullish bias. Upper band touches are continuation signals, not reversal signals. Lower band touches in this regime are buying opportunities.
• Middle band falling clearly: bearish bias. The opposite of the above.
• Middle band flat: range-bound. Now the textbook mean-reversion logic applies.

Skipping that read is what produces the losing trades. The bands themselves are fine. The trader skipping context is the problem.

Bollinger Bands vs other volatility indicators

vs Keltner Channels. Keltner Channels use ATR (Average True Range) instead of standard deviation, and an EMA instead of an SMA for the middle line. Keltner is smoother and reacts less violently to single-bar shocks. Bollinger widens faster on sudden volatility spikes; Keltner adapts more gradually. Some traders pair them: a Keltner inside a Bollinger creates the "squeeze indicator" - when Bollinger contracts inside Keltner, the market is unusually compressed.

vs ATR. ATR is a single line that measures average bar range. Bollinger Bands are an envelope around price. They measure related but distinct things: ATR is the "size of a typical move"; Bollinger Bands are "where price is relative to its recent statistical range." ATR is used for stop sizing; Bollinger Bands are used for context and entries.

vs Donchian Channels. Donchian Channels are simply the highest high and lowest low of the last N periods. They are a pure breakout indicator with no smoothing. Bollinger Bands are statistical and adaptive; Donchian is hard-edged. Trend-following systems often use Donchian; mean-reversion and squeeze systems use Bollinger.

Common questions

Should I change the 20-period, 2-standard-deviation default? Almost never. John Bollinger spent years testing alternatives and the 20/2 settings remain his preferred defaults. Shorter periods (10/2) are noisier; longer periods (50/2) lag the trend. Wider deviations (20/2.5) catch fewer breakouts; narrower (20/1.5) catch too many false ones. The default is the equilibrium.

Does the squeeze always lead to a breakout? Not always, but almost always within a reasonable window. Volatility is mean-reverting - low-volatility regimes do not last indefinitely. What is not predictable is the direction. A squeeze tells you "a move is coming"; it does not tell you which way.

Can Bollinger Bands work on logarithmic charts? The math is computed on closing prices, not on the chart scale. Switching between linear and log changes the visual appearance but not the indicator readings.

What is %B? %B (percent B) is a derivative that plots where price sits relative to the bands as a single oscillator. %B = (price - lower band) / (upper band - lower band). It reads 0 at the lower band, 1.0 at the upper band, 0.5 at the middle. Useful for divergence reads and for comparing band positioning across instruments.

Why do prices sometimes spend long stretches outside the bands? Because financial markets are not normally distributed. The "95 percent containment" math assumes Gaussian returns; real returns have fat tails, especially during news, earnings, and macro events. The bands are a useful approximation, not a hard boundary.

Should I trade Bollinger Bands without other indicators? Trying to trade purely on band touches is what produces the losing-trader cliché. Combine with at least one trend filter (the slope of the middle band counts) and one momentum check (RSI divergence, MACD position relative to zero, or simply structure).

When to use Bollinger Bands and when not to

Use Bollinger Bands when:

• You need to know whether the current volatility regime is compressed or expanded
• You are setting up for a squeeze breakout on a clean range
• You are checking whether a trending move is showing band-ride strength

Skip Bollinger Bands when:

• You are using them as a directional signal (the touch-and-fade approach) - that is what loses money
• The instrument has chronically erratic volatility (very thin small-caps, low-liquidity altcoins) - σ is unreliable in those markets
• You already have ATR and Keltner Channels on the chart - more envelopes do not add information