Bollinger Breakouts

How to read squeezes and breakouts

Bands and bandwidth

Bollinger Bands use a 20-period simple moving average and an upper and lower boundary at two standard deviations. Bandwidth, BBW, is the upper-minus-lower distance divided by the middle average. When the bands narrow, BBW falls; when they widen, BBW rises. Percent B describes where the latest close sits in the channel. The calculation uses only finished candles to avoid a transient live price being mistaken for a confirmed exit. A close beyond a band is an observation about one candle, not evidence that subsequent candles will keep travelling in the same direction.

Intensity and history

Before any breakout can be recorded, the asset must first meet the squeeze rule. Its current BBW is compared with the previous 250 valid widths on the same timeframe. An intensity at or above 90 places width in roughly the narrowest tenth of this own-history window. This normalization matters because different assets have very different usual ranges. At least 20 bars are needed for the initial bands and 250 earlier width readings for the comparison. If source coverage is inadequate, the asset is excluded rather than shown with a misleading intensity.

From squeeze to breakout

A confirmed event follows at least two squeeze bars. Within the first three bars after leaving squeeze, BBW must rise by at least 25% from the period minimum and the candle must close outside the upper or lower band. The direction records which band was crossed. An expanding BBW without an outside close does not qualify. A band touch before the candle closes does not qualify either. Tightening and in-squeeze stages are still visible on the companion page, allowing you to trace the full sequence from contraction to observed exit.

Keltner and records

Keltner confirmation means Bollinger Bands fit inside an EMA20 channel widened by 1.5 average true ranges during compression. This TTM-style subtype can help identify stronger compression, but it does not determine which side will break. A year or all-history record refers to the narrowest width in a completely covered reference window; available source history is named explicitly. These badges add context to an event. They are not a quality guarantee, and they should not replace inspection of the actual price and liquidity of the asset.

Directional tilt

The squeeze page’s directional tilt combines Percent B, MACD histogram and slope, the higher-timeframe trend, the balance of rising and falling volume, and earlier breakout directions for the same asset. It is a descriptive label with an evidence breakdown. It cannot turn a squeeze into a directional forecast. Once an outside close is recorded, the breakout page shows the observed direction separately from any earlier tilt. Comparing the two may be useful for research, but a single observation does not validate a trading rule.

Outcomes and false exits

The outcome section waits for later closed bars before reporting three, ten and twenty bar returns. It also tracks the maximum favourable and adverse move inside each completed horizon. If any close returns inside the channel within the following three bars, the event is marked as a false breakout. False does not mean the data were wrong; it means the apparent exit did not persist under this explicit rule. Past sample counts and median returns vary with market conditions and history depth. They cannot tell you what the next breakout will earn. This page is market analysis, not a recommendation to buy or sell.