Marubozu

Single-bar strong-continuation candle with body equal to range and (almost) no shadows. A clean directional bar — one side won the entire bar with no contest. Signals strong momentum in the direction of the body.

Quick reference

Item	Value
Family	Candlestick Patterns
Input type	Candle
Output type	f64 — +1.0 bullish, -1.0 bearish, 0.0 otherwise
Output range	{-1.0, 0.0, +1.0}
Default parameters	shadow_tolerance = 0.05 (5% of range per shadow)
Warmup period	1
Interpretation	Strong momentum confirmation in the direction of the body

Formula

range        = high - low
upper_shadow = high - max(open, close)
lower_shadow = min(open, close) - low

shadows OK = (upper_shadow <= tol · range)
          AND (lower_shadow <= tol · range)

bullish (+1.0): shadows OK AND close > open
bearish (-1.0): shadows OK AND close < open
otherwise: 0.0

shadow_tolerance defaults to 0.05 (5% of bar range allowed on each side). See crates/wickra-core/src/indicators/marubozu.rs.

Parameters

Name	Type	Default	Constraint	Description
shadow_tolerance	f64	0.05	finite, [0, 1)	Max allowed shadow as fraction of range.

Marubozu::new returns Error::InvalidPeriod for out-of-range shadow_tolerance. Marubozu::default() returns the 0.05 factory.

Signed ±1 encoding

This pattern already emits the uniform candlestick sign convention shared across the family — +1.0 bullish, −1.0 bearish, 0.0 no pattern — so it drops straight into a machine-learning feature matrix where the bullish and bearish variants of the pattern occupy a single dimension.

Inputs / Outputs

Indicator<Input = Candle, Output = f64>. Python: Marubozu(tol).batch(open, high, low, close) returns a 1-D np.ndarray. Node: same; update(candle) returns number | null.

Warmup

warmup_period() == 1. Stateless.

Edge cases

• Zero-body bar. Doji-style bars have body = 0; the > / < strict inequality fails, output is 0.0.
• Tolerance choice. 0.0 (strict zero shadows) is rarely satisfied by real data; 0.05 (5%) is a practical default.
• Reset. Stateless.

Examples

Rust

use wickra::{Candle, Indicator, Marubozu};

fn main() -> Result<(), Box<dyn std::error::Error>> {
    // Bullish Marubozu: open=100, close=110, high=110.05, low=99.95
    let c = Candle::new(100.0, 110.05, 99.95, 110.0, 1.0, 0)?;
    let mut m = Marubozu::default();
    println!("{:?}", m.update(c));  // +1.0
    Ok(())
}

Python

import numpy as np
import wickra as ta

o = np.array([100.0])
h = np.array([110.05])
l = np.array([99.95])
c = np.array([110.0])

m = ta.Marubozu(0.05)
print(m.batch(o, h, l, c))

Node

const wickra = require('wickra');
const m = new wickra.Marubozu(0.05);
console.log(m.batch([100], [110.05], [99.95], [110]));

Streaming

use wickra::{Candle, Indicator, Marubozu};

let mut m = Marubozu::default();
let candle_stream: Vec<wickra::Candle> = Vec::new(); // your live OHLCV candle feed
for bar in candle_stream {
    if m.update(bar) == Some(1.0) { /* bullish continuation */ }
    if m.update(bar) == Some(-1.0) { /* bearish continuation */ }
}

Interpretation

• Strong momentum. No shadow = no opposing pressure. A Marubozu in the direction of the prevailing trend confirms strong trend continuation.
• At range extremes. A Marubozu against a range is often the first bar of a range breakout. A Marubozu in the trend's direction near a swing high may also signal exhaustion (after a long run).
• Pair with volume. A Marubozu on high volume is significantly more meaningful than one on low volume.

Common pitfalls

• Tolerance too tight. With shadow_tolerance = 0.0, almost no real bar qualifies. Default 0.05 is the practical floor.
• Treating every Marubozu as continuation. Late-stage Marubozu (after extended trend) often marks exhaustion, not continuation. Pair with trend duration / divergence indicators.

References

• Steve Nison, Japanese Candlestick Charting Techniques (1991) — Marubozu among the basic candle definitions.

See also

• ThreeSoldiersOrCrows — multi-bar continuation pattern.
• Engulfing — full-body engulfing reversal.
• SpinningTop — Marubozu's opposite (long shadows, small body).
• Indicators-Overview — full taxonomy.