BomarBands

Adaptive percentage bands around a moving average, sized so that a fixed coverage fraction of recent closes falls inside them. The pre-Bollinger percentage band.

Quick reference

Field	Value
Family	Bands & Channels
Input type	f64
Output type	BomarBandsOutput { upper, middle, lower }
Output range	unbounded; lower ≤ middle ≤ upper
Default parameters	period = 20, coverage = 0.85
Warmup period	period (exact — first emission on bar period)
Interpretation	Self-tuning percentage envelope: widens in turbulent regimes, tightens in quiet ones, with no volatility input.

Formula

middle = SMA(close, period)
dev_i  = | close_i / middle − 1 |             relative distance from the midline
p      = coverage-quantile of { dev_i }        type-7 interpolation
upper  = middle + |middle| · p
lower  = middle − |middle| · p

John Bollinger credits the Bomar Bands as an inspiration for his own work: percentage bands around a moving average, with the percentage tuned so that a fixed share (classically ~85%) of price stayed within. Wickra realises that idea deterministically — the half-width is the coverage quantile of the relative deviations from the midline, so by construction coverage of the window's closes lie inside the bands. Source: crates/wickra-core/src/indicators/bomar_bands.rs.

Parameters

Name	Type	Default	Constraint	Source
period	usize	20	>= 1	BomarBands::new (bomar_bands.rs:76)
coverage	f64	0.85	finite, in (0.0, 1.0]	bomar_bands.rs:76

period == 0 returns [Error::PeriodZero]; a coverage outside (0.0, 1.0] or non-finite returns [Error::InvalidParameter]. Python defaults come from #[pyo3(signature = (period=20, coverage=0.85))]; the Node constructor takes both arguments explicitly.

Inputs / Outputs

use wickra::{Indicator, BomarBands, BomarBandsOutput};
// BomarBands: Input = f64, Output = BomarBandsOutput
const _: fn(&mut BomarBands, f64) -> Option<BomarBandsOutput> =
    <BomarBands as Indicator>::update;

• Python streaming. update(value) returns (upper, middle, lower) or None.
• Python batch. BomarBands.batch(prices) returns an (n, 3) np.ndarray with columns [upper, middle, lower]; warmup rows are NaN.
• Node streaming. update(value) returns a { upper, middle, lower } object or null.
• Node batch. batch(prices) returns a flat Array<number> of length n * 3 interleaved per row [u0, m0, l0, …].

Warmup

warmup_period() reports period and is exact: the SMA midline needs a full window before the bands are defined, so the first non-None output lands on bar period. Pinned by warms_up_then_emits (three Nones then Some for period = 4).

Edge cases

• Zero midline. When the window mean is exactly 0 the relative deviation is undefined; the bands collapse onto the midline (upper == middle == lower) — pinned by zero_midline_collapses_bands (window [3, −3]).
• coverage = 1.0. The half-width becomes the maximum relative deviation, so every close in the window lies inside the bands.
• Ordering. upper ≥ middle ≥ lower always holds because offset ≥ 0.
• Reset. reset() clears the window and scratch buffer; the next update restarts warmup (test reset_clears_state).

Examples

Rust

use wickra::{BatchExt, BomarBands, Indicator};

fn main() -> Result<(), Box<dyn std::error::Error>> {
    let mut bb = BomarBands::new(4, 0.85)?;
    for v in bb.batch(&[100.0, 102.0, 98.0, 104.0]) {
        println!("{:?}", v);
    }
    Ok(())
}

Output:

None
None
None
Some(BomarBandsOutput { upper: 104.0, middle: 101.0, lower: 98.0 })

SMA = 101; the relative deviations are {1, 1, 3, 3} / 101; the 0.85 quantile is 3/101, so offset = 101 · 3/101 = 3 and the bands sit at 101 ± 3. This matches the known_bands test.

Python

import numpy as np
import wickra as ta

bb = ta.BomarBands(4, 0.85)
print(bb.batch(np.array([100.0, 102.0, 98.0, 104.0])))

Output:

[[ nan  nan  nan]
 [ nan  nan  nan]
 [ nan  nan  nan]
 [104. 101.  98.]]

Node

const ta = require('wickra');
const bb = new ta.BomarBands(4, 0.85);
bb.update(100); bb.update(102); bb.update(98);
console.log(bb.update(104)); // { upper: 104, middle: 101, lower: 98 }

Interpretation

Bomar Bands are a self-calibrating percentage envelope:

1. Containment. By design ~coverage of recent closes sit inside the bands, so a close outside is a genuine (1 − coverage)-tail event for the current regime — a cleaner overbought/oversold trigger than a fixed-percent MaEnvelope.
2. Adaptive width. The bands breathe with realised dispersion without an ATR or sigma input; a sudden widening flags a volatility regime change.

Common pitfalls

• Reading the width as volatility. The half-width is a quantile of deviations, not a standard deviation — it is bounded by the largest deviation in the window and ignores the extreme tail beyond the coverage rank.
• Non-positive series. The percentage construction assumes a meaningful non-zero midline; on series that cross zero (e.g. spreads) the midline can vanish and the bands collapse.

References

• Bollinger, John, Bollinger on Bollinger Bands, McGraw-Hill, 2001 (historical note crediting the Bomar percentage bands).

See also

• MaEnvelope — fixed-percent band around an MA.
• BollingerBands — sigma-based dispersion envelope.
• QuartileBands — non-parametric quartile envelope.
• PercentB — close position inside Bollinger Bands.