HikkakeModified

Dan Chesler's Modified Hikkake (TA-Lib CDLHIKKAKEMOD) — a refinement of the Hikkake trap. Two nested inside bars coil the market, the first inside bar already closes near the extreme the false break will run towards, and the fourth bar breaks out of the second inside bar the wrong way.

Quick reference

Field	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	none — HikkakeModified::new()
Warmup period	4 (first three bars return None)
Interpretation	Double-inside-bar false-breakout trap

Formula

bar2 inside bar1 : high2 < high1  &&  low2 > low1
bar3 inside bar2 : high3 < high2  &&  low3 > low2
bullish (+1.0): bar4 makes a lower high AND lower low than bar3,
                and bar2 closed near its low   (close2 <= low2  + 0.2 · range2)
bearish (-1.0): bar4 makes a higher high AND higher low than bar3,
                and bar2 closed near its high  (close2 >= high2 − 0.2 · range2)
range2 = high2 − low2

Compared with the plain Hikkake (one inside bar, then a break), the modified pattern needs two nested inside bars and a close filter on the first inside bar: for a bullish setup it must already have closed in the bottom fifth of its range, for a bearish one in the top fifth. "Near" is a fixed fifth of bar 2's range — TA-Lib's Near factor of 0.2 applied geometrically rather than to a rolling average of candle ranges. The signal fires on the breakout bar 4; the later confirmation bar is not flagged separately. Pattern-shape check only — no trend filter is applied. See crates/wickra-core/src/indicators/hikkake_modified.rs.

TA-Lib parity. On the TA-Lib reference pattern series the signals on the breakout bar (bar 4, the pattern's setup bar) are identical to CDLHIKKAKEMOD (±1 here, ±100 there). TA-Lib additionally emits ±200 on a later confirmation bar; Wickra flags only the setup bar. TA-Lib sizes bodies and shadows against 5- or 10-bar rolling averages ("candle settings"), while Wickra sizes them against the pattern's own bars.

Parameters

None. Constructed with HikkakeModified::new().

Signed ±1 encoding

Emits the uniform candlestick sign convention — +1.0 bullish, −1.0 bearish, 0.0 no pattern — a single feature-matrix dimension.

Inputs / Outputs

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

• Warmup returns None. The first three bars return None; from bar 4 on every bar emits +1.0, -1.0 or 0.0.
• Node. update(open, high, low, close) → number | null; batch(open, high, low, close) → Array<number> (NaN on warmup).
• Python. update(candle) → float | None; batch(open, high, low, close) → 1-D array.array('d') (NaN on warmup, 0.0 on no-match).

Warmup

warmup_period() == 4. The four-bar window must fill, so the first three bars return None (first_three_bars_return_none, accessors_and_metadata).

Edge cases

• Bar 2 must close near the extreme. A first inside bar that closes mid-range fails the close filter and yields 0.0 (second_bar_close_not_near_extreme_yields_zero).
• Two nested inside bars. If bar 3 is not inside bar 2 (or bar 2 not inside bar 1) the result is 0.0 (not_double_inside_bar_yields_zero).
• Bearish mirror. Higher high and higher low on bar 4 with bar 2 closing in the top fifth gives -1.0 (bearish_modified_hikkake_is_minus_one).
• Reset. reset() clears the three-bar cache (reset_clears_state).

Examples

Rust

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

fn main() -> Result<(), Box<dyn std::error::Error>> {
    let mut t = HikkakeModified::new();
    println!("{:?}", t.update(Candle::new(10.0, 15.0, 5.0, 12.0, 1.0, 0)?)); // wide bar
    println!("{:?}", t.update(Candle::new(11.0, 13.0, 7.0, 7.5, 1.0, 1)?));  // inside bar, closes near low
    println!("{:?}", t.update(Candle::new(9.0, 12.0, 8.0, 10.0, 1.0, 2)?));  // second inside bar
    println!("{:?}", t.update(Candle::new(9.0, 11.0, 6.0, 9.0, 1.0, 3)?));   // lower high + lower low
    Ok(())
}

Output:

None
None
None
Some(1.0)

Bar 2 (13 / 7) sits inside bar 1 (15 / 5) and closes at 7.5, within 0.2 · 6 = 1.2 of its low; bar 3 (12 / 8) sits inside bar 2. Bar 4 makes a lower high (11 < 12) and lower low (6 < 8) than bar 3 — a bullish modified hikkake. This matches bullish_modified_hikkake_is_plus_one.

Python

import numpy as np
import wickra as ta

o = np.array([10.0, 11.0,  9.0, 9.0])
h = np.array([15.0, 13.0, 12.0, 11.0])
l = np.array([5.0,  7.0,  8.0,  6.0])
c = np.array([12.0, 7.5, 10.0,  9.0])

print(ta.HikkakeModified().batch(o, h, l, c))  # array('d', [nan, nan, nan, 1.0])

Node

const ta = require('wickra');
const t = new ta.HikkakeModified();
t.update(10, 15, 5, 12);  // null
t.update(11, 13, 7, 7.5); // null
t.update(9, 12, 8, 10);   // null
console.log(t.update(9, 11, 6, 9)); // 1

Streaming

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

let mut t = HikkakeModified::new();
let candle_stream: Vec<wickra::Candle> = Vec::new(); // your live OHLCV candle feed
for bar in candle_stream {
    match t.update(bar) {
        Some(1.0)  => { /* false downside break — lean long */ }
        Some(-1.0) => { /* false upside break — lean short */ }
        _ => {}
    }
}

Interpretation

1. Stricter than plain hikkake. Two nested inside bars plus the bar-2 close filter make the setup rarer than the single-inside-bar Hikkake — fewer, more selective signals.
2. Fade the trap. Lean in the direction opposite the bar-4 break: a downside break after bar 2 closed near its low is read bullish, and vice versa.
3. Range markets. Like the plain hikkake, most useful in choppy ranges.

Common pitfalls

• No confirmation. The signal fires on bar 4 itself; whether the break actually fails (price re-crossing the inside bars) is not checked — confirm it yourself on subsequent bars.
• Confusing with plain hikkake. Hikkake needs one inside bar and the high/low break; this one needs two nested inside bars and the bar-2 close filter.
• Near is geometric. The 0.2 · range2 threshold is fixed; TA-Lib's version scales Near by a rolling average of candle ranges, so results can differ on the close filter.

References

• Daniel Chesler, "Trading False Moves with the Hikkake Pattern", Active Trader (2004) — the original hikkake and its modified variant.
• TA-Lib, CDLHIKKAKEMOD (Modified Hikkake Pattern).
• Steve Nison, Japanese Candlestick Charting Techniques (1991).

See also

• Hikkake — the plain break-only variant.
• Harami — the inside-bar reversal pattern.
• Indicators-Overview — the full taxonomy.