Bollinger Bands: the squeeze, the bounce, and what actually works

Bollinger Bands are three lines: a moving average in the middle and a band above and below it. The retail reading — sell at the upper band, buy at the lower — is wrong often enough to drain an account. The bands do not mark expensive and cheap. They mark volatility. Get that straight and the indicator changes from a contrarian trap into a genuine tool.

How the bands are built

The middle line is a simple moving average, usually 20 periods. The outer bands sit a number of standard deviations away — typically two — measured over the same window. Standard deviation is a volatility measure, so the bands widen when the market gets volatile and contract when it goes quiet. The band width is the signal; the band touch usually is not.

Statistically, with 2-standard-deviation bands, roughly 95% of recent prices fall inside them — but only if returns were normally distributed, which crypto returns emphatically are not. Crypto has fat tails: price rides the band and pierces it far more often than the textbook says. That alone kills the naive 'reversion at the band' rule.

The squeeze — the setup that holds up

The most robust Bollinger pattern is the squeeze: the bands contract to an unusually narrow width, signalling that volatility has collapsed. Volatility is mean-reverting — quiet periods are followed by active ones — so a tight squeeze is a reliable tell that a larger move is coming. What it does not tell you is direction. The squeeze is a 'something is about to happen' signal; you need a separate trigger (a breakout, a trend filter) to decide which way to trade it.

The bounce — only with a regime filter

The mean-reversion 'bounce' off a band can work — but only when the market is ranging. In a trend, price walks the band and every counter-trend bounce trade is a loss. So the bounce is never a standalone rule. Gate it: only take band-touch reversion entries when a trend filter says the market is rangebound (flat long EMA, low ADX). Inside a range the bounce has an edge; outside one it is a slow bleed.

Backtest both modes — separately

Squeeze-breakout and band-bounce are opposite strategies — one bets on expansion, one on reversion — and they win in opposite regimes. Backtesting them blended together produces a mush that hides which one is actually working. Test each on its own, with its regime filter, and you will see clearly: the breakout earns in trends, the bounce earns in ranges. Then, if you want, combine them with the regime filter as the switch.

Bollinger Bands are in Noon Barbari's built-in indicators, and a BB squeeze strategy is one of the reference templates. Build the squeeze and the bounce as two rule sets, backtest each separately, and let the numbers show you which regime each one needs.

Bollinger Bands answer one question well: how volatile is this market right now, relative to its recent self? Use them for that — the squeeze for timing, the band width for regime — and stop using a band touch as a price verdict it was never designed to give.