Bollinger Squeeze Release: Trade the Expansion

Volatility mean-reverts even when price does not. A stretch of unusually narrow Bollinger bandwidth - a squeeze - is normally resolved by an expansion, and that expansion tends to run in whichever direction price first breaks out of the compressed band. This strategy waits for the compression, then trades the release rather than trying to predict it. Four things have to line up. Bandwidth on a 34-period band must have sat below 95% of its own slow baseline at some point in the last 11 candles; bandwidth must be expanding again on the current candle; price must have closed outside the band; and the breaking candle itself must point the same way as the break. That last condition is what keeps the strategy out of wick-driven band pokes that have already reversed by the close. A 66-period trend EMA gates each side, so a lower-band break is only shortable below trend and an upper-band break is only buyable above it. Both directions are live, and over the tested window both were profitable independently: 97 short trades returned 0.77% and 84 long trades returned 0.18%. The exit is deliberately loose, because a released move needs room. A position is held until price crosses back over the trend EMA - the thesis is dead - or until bandwidth falls back to its own baseline, meaning the expansion being traded is over. An earlier version of this design exited at the band midline instead, and that failed badly for a specific and instructive reason: right after a squeeze the bands are narrow, so the midline sits only a fraction of a percent from the entry and acts as a hair-trigger stop. It cut the median trade at a loss after nine hours and turned the whole strategy negative. Replacing it with the trend-and-bandwidth exit is what made the technique work. Risk is capped hard: every position carries a -10% stoploss and a flat 10% take-profit, and a hard stop is not optional because the short side has unbounded loss if price rallies against it. Maximum drawdown over the tested window was 0.51%. Every signal parameter is exposed as an IntParameter or DecimalParameter. The published defaults are the winning values from a 100-epoch hyperopt run over the buy space, executed against the in-sample leg only and never against the out-of-sample data. Two of the four settled at the edge of their search range, so a wider re-optimisation on your own pairs may find better ground. Tested on six Binance USDT-margined perpetual futures - BTC, ETH, ATOM, DOT, LINK and XRP - on the 1h timeframe in isolated-margin mode. Past backtest performance is not a prediction of future returns. This window was a severe bear market; the design has not been validated across a sustained bull market.