Bollinger Bands Trading Strategy: Separate the Bands from the Rules
Bollinger Bands describe price relative to a moving average and measured dispersion. Turning that description into a strategy requires separate rules for entries, exits, sizing and timing. A touch of an outer band is not sufficient evidence of a reversal, and a narrow band does not determine the direction or date of a future move.
Key Takeaways
- The familiar 20-period, two-deviation configuration is a default, not a universal optimum.
- Band multiplier changes outer-band distance while leaving an unchanged basis calculation unchanged.
- Reversion and continuation are different hypotheses and need separate tests.
- Adding RSI or volume does not automatically improve accuracy.
Check the calculation and units
In a traditional configuration, the middle band is an SMA of the selected price source over n bars. The upper and lower bands are that basis plus or minus k times the standard deviation over the chosen window. Record the source, length, averaging method and deviation convention when comparing implementations.
- Hypothetical basis 100, standard deviation 2 and multiplier 2 give bands at 104 and 96.
- Changing only the multiplier to 1.5 gives 103 and 97. The middle remains 100, so its crossing events do not move merely because the outer bands narrow.
- Twenty periods means twenty bars of the selected calculation timeframe, not necessarily twenty days.
- Two standard deviations do not imply a guaranteed 95% probability of containment or reversal.
Specify a reversion hypothesis
For an untested example, define a long event as a completed close returning above its current lower band after the prior close was below its prior lower band. Decide whether to exit at the middle band, after a fixed number of bars or under another explicit rule. This is different from buying every lower-band touch.
- Set the position limit, entry timing and loss-control policy before evaluation.
- If RSI is added, specify its period, threshold and whether a crossing or a level is required.
- Compare the filtered and unfiltered rules on the same separate evaluation interval.
- Inspect persistent declines and gaps; the moving reference does not require price to recover.
Measure contraction before testing a squeeze breakout
BandWidth is commonly expressed as (upper − lower) / middle, sometimes multiplied by 100. State the convention. A squeeze needs a defined relative threshold and lookback; the word “narrow” alone is not reproducible. Then define the event that ends the setup and supplies a direction.
- For a research specification, a prior completed-bar BandWidth below a chosen historical percentile can arm a setup, followed by a separately defined close beyond an outer band. Those choices are untested parameters.
- Use only observations available at the decision time when calculating the percentile.
- A contraction can persist and an initial break can reverse. There is no established 50/50 outcome or guaranteed volume-filter benefit here.
Keep a band walk distinct from a reversal
Price can remain near or beyond one outer band during a trend. A continuation study might require a predefined sequence of upper-band closes and then a pullback event. Record the exact sequence, trend condition and expiry of the setup rather than labeling a winning move a band walk afterward.
- Do not combine a reversion entry and continuation interpretation without stating which rule has priority.
- Changing the multiplier may change band-walk qualification, but it does not directly alter the middle average.
- Choose an exit independently; the band itself is not an order or a guaranteed loss limit.
Evaluate settings without asset-class shortcuts
A more volatile price series already changes the measured standard deviation. That alone does not prove that crypto requires a 2.5–3 multiplier or forex requires a shorter length. Compare a limited, predeclared set of candidates using consistent data, execution costs and later evaluation dates.
- Pineify can draft Pine source from an explicit band rule; inspect and compile it in TradingView.
- The Strategy Optimizer requires Advanced or above and searches exposed inputs, not arbitrary new confirmation logic unless the script implements it.
- Inspect individual trades and changing realtime values as well as summary metrics.
- Keep all variants and costs with the results; optimization does not establish a permanently optimal setting.
Sources and verification
Last verified:
Educational information, not investment advice or a record of tested returns. Examples illustrate mechanics. Trading can lose money, and historical or simulated results do not guarantee future performance.