Bollinger Bands for Forex: Volatility, Breakouts and Pullbacks

What the bands actually measure

Bollinger Bands place an upper and lower envelope around a moving average. The distance between them expands when the closing prices in the calculation window are more dispersed, and contracts when they cluster together. That makes the indicator useful for comparing today’s price behaviour with its own recent history.

Two questions must stay separate: is volatility contracting or expanding, and is price trending or rotating inside a range? A narrow envelope answers the first question. It does not tell you which direction the next move will take. A price close near the upper band can reflect persistent upward pressure; treating every such close as a reason to sell can repeatedly position you against a trend.

John Bollinger specifically cautions that touching a band is insufficient as a trading signal and that price can travel along a band during a trend. The familiar settings are starting parameters, not an assurance of an edge. Read the creator’s rules.

Calculation, settings and two useful measurements

For the standard close-price version, let M be the simple average of the last N closes and let σ be their population standard deviation: the square root of the average squared distance from M. Then upper band U = M + kσ and lower band L = M − kσ. A common starting point is N = 20 and k = 2, with chart shift set to zero. MetaQuotes documents this calculation.

Two derived quantities make comparisons easier: %B = (close − L)/(U − L), and BandWidth = (U − L)/M × 100. %B equals 0 at the lower band and 1 at the upper band; values outside that interval are possible. Avoid division when the bands have zero width. BandWidth expresses width as a percentage of the middle line.

These are descriptions of the observed price window. Two standard deviations do not create a guaranteed 95% containment interval for future forex prices, and BandWidth is not the same calculation as ATR.

Choose a regime before choosing a signal

A breakout approach looks for compressed volatility followed by directional expansion. A range approach looks for a failed excursion followed by a return toward the centre. Both can produce visually attractive examples, but they make opposite assumptions. Write down which interpretation applies before the next candle appears.

For a range study, mark horizontal boundaries from previously available data and require a close back inside the band before considering a reversal. For a breakout study, look for a close outside a boundary and accept that the entry will come after some movement has already occurred. Do not switch from breakout to mean reversion simply because the first trade loses. That changes the strategy in response to its outcome.

Extra indicators should answer a different question. Adding several oscillators calculated from the same closing prices rarely supplies three independent confirmations.

A reproducible H1 breakout study

The following is an educational research specification, not a tested recommendation. Use only completed EUR/USD hourly candles, Bollinger Bands (20, 2), and a 50-period EMA of closes.

1. At the close of bar t−1, define a squeeze when its BandWidth is no higher than the tenth percentile of the 120 BandWidth readings ending at t−2. Use the nearest-rank percentile consistently.
2. At the close of bar t, require close(t) above its upper band, BandWidth(t) above BandWidth(t−1), and EMA50(t) above EMA50(t−5). This is the long trigger.
3. Enter at the next available ask after that close. Skip the entry if the spread exceeds 2 pips or the fill is more than 0.2 ATR(14) above the signal close.
4. Place the initial stop below the lowest low of bars t−4 through t, with a further 0.25 ATR(14) buffer. Fix that stop after entry. Size the position so the planned loss, including estimated costs, is no more than 0.5% of equity.
5. Set a target at twice the initial price risk. Exit earlier at the next available bid after a completed candle closes below its middle band, or after 24 completed candles. Use whichever exit occurs first and allow only one position.

Mirror every inequality and bid/ask convention to test shorts separately. The thresholds above are explicit design choices that require validation.

Worked EUR/USD example

Assume the signal candle finishes with a middle band of 1.0850 and σ of 0.0010. The bands are therefore 1.0870 and 1.0830. A close at 1.0874 gives %B = 1.10 and BandWidth of approximately 0.369%. The close is outside the band; it does not by itself prove that a breakout will continue.

Suppose the other study conditions also pass, the actual long fill is 1.0876, and the stop calculation produces 1.0836. Initial price risk is 40 pips, so a 2R target is 1.0956. On a hypothetical USD 10,000 account, a 0.5% planned risk budget is USD 50. If one standard EUR/USD lot is worth USD 10 per pip, the size before costs is 50/(40 × 10) = 0.125 lots. Reduce and round down for commission, an execution allowance and the broker’s volume step.

This illustrates the arithmetic only. If price immediately reverses, the same properly formed signal can produce a loss.

Failure modes and a realistic test

A squeeze can last longer than expected, break in both directions, or end in a news gap. A slow trend can repeatedly touch a band without reverting. A jump also changes the average and deviation used to draw the envelope, so a visually changing band is not an independent market event.

Test the fixed rules on a chronological development sample, then an untouched later period. Include trending, quiet and volatile conditions; report trade count, net expectancy, maximum drawdown and the longest losing sequence. Check nearby settings rather than selecting one isolated optimum. A result that collapses when the period changes from 20 to 19 deserves caution.

Model historical bid/ask spreads, commission, slippage and overnight financing. When a stop and target fall inside the same candle, use suitable tick data or a conservative ordering rule. Do not assume the target was hit first. Predefine any event blackout, such as no new entries from 30 minutes before until 30 minutes after a scheduled policy decision.

Set up the study in MT4 or MT5

On desktop, add Bollinger Bands from the Indicators menu or Navigator. Enter period 20, deviation 2, shift 0 and Apply to Close. Add EMA50 separately and save the arrangement as a template. Use the Data Window to inspect a completed bar’s exact values instead of reading approximate positions from the chart.

In an automated study, current bar 0 is still forming; the last completed candle is bar 1. Compare the completed values, then execute on the next available quote. %B and percentile-based BandWidth conditions require calculation or a suitable custom indicator. Check that any custom version uses the same standard deviation convention as your chart. Keep a journal screenshot at signal time so later redraws cannot alter the recorded decision.