WaveTrend

LazyBear's Wave Trend Oscillator — a two-line mean-reverting momentum gauge built from the typical price and three cascaded EMAs. Most useful at extremes, where wt1 crossing back through wt2 flags reversals.

Quick reference

Field	Value
Family	Trend & Directional
Input type	Candle (uses high, low, close to form the typical price ap)
Output type	WaveTrendOutput { wt1, wt2 }
Output range	unbounded; typically [−100, +100] in practice, extremes near ±60
Default parameters	channel_period = 10, average_period = 21, signal_period = 4
Warmup period	2 · channel + average + signal − 3 (42 for classic)
Interpretation	wt1 crossing above wt2 from < −60 is bullish; crossing below from > +60 is bearish.

Formula

ap_t  = (high_t + low_t + close_t) / 3
esa_t = EMA(ap, channel)
d_t   = EMA(|ap − esa|, channel)
ci_t  = (ap_t − esa_t) / (0.015 · d_t)
wt1_t = EMA(ci, average)
wt2_t = SMA(wt1, signal)

On a perfectly flat market the SMA-seeded EMA introduces a single-ULP drift between ap and esa, which would otherwise make the ratio explode to −1/0.015 = −66.67. The implementation guards with a price-scaled flat tolerance (d <= 16 · EPSILON · max(|esa|, 1) ⇒ ci := 0), so a fully motionless market reports (0, 0).

Parameters

Name	Type	Default	Constraint	Source
channel_period	usize	10	>= 1	WaveTrend::new (wave_trend.rs:77)
average_period	usize	21	>= 1	wave_trend.rs:77
signal_period	usize	4	>= 1	wave_trend.rs:77

Any zero period returns [Error::PeriodZero]. WaveTrend::classic() returns (10, 21, 4) (as a Result). Python defaults come from the pyo3 signature; the Node constructor takes all three explicitly. The public class is WaveTrend in both bindings.

Inputs / Outputs

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

• Python. update(candle) returns (wt1, wt2) or None; batch(high, low, close) returns an (n, 2) np.ndarray with columns [wt1, wt2]; warmup rows are NaN.
• Node. update(high, low, close) returns a { wt1, wt2 } object or null; batch(high, low, close) returns a flat Array<number> of length 2n, interleaved [wt1_0, wt2_0, …].

Warmup

warmup_period() returns 2 · channel + average + signal − 3. esa emits at channel inputs; the deviation EMA needs channel of its own values (ready at 2·channel − 1); wt1's EMA adds average − 1; the wt2 SMA adds signal − 1. For the classic config that is 2·10 + 21 + 4 − 3 = 42. Pinned by first_emission_at_warmup_period.

Edge cases

• Flat market. The flat-tolerance guard collapses ci to 0, so both lines stay at 0 (test constant_series_yields_zero_lines).
• Pure trend. A clean uptrend drives both lines positive (test pure_uptrend_is_positive); the downtrend mirror is negative.
• Always finite. Both lines stay finite across volatile inputs (test outputs_remain_finite).
• Reset. reset() resets all three EMAs and the signal SMA.

Examples

Rust

use wickra::{BatchExt, Candle, Indicator, WaveTrend};

fn main() -> Result<(), Box<dyn std::error::Error>> {
    let candles: Vec<Candle> = (0..120)
        .map(|i| {
            let base = 100.0 + f64::from(i) * 0.5;
            Candle::new(base, base + 1.0, base - 0.5, base + 0.5, 1.0, i64::from(i)).unwrap()
        })
        .collect();
    let mut w = WaveTrend::classic()?; // (10, 21, 4)
    if let Some(o) = w.batch(&candles).into_iter().flatten().last() {
        println!("wt1={:.3} wt2={:.3}", o.wt1, o.wt2);
    }
    Ok(())
}

Python

import wickra as ta
w = ta.WaveTrend(10, 21, 4)
out = w.batch(high, low, close)  # shape (n, 2): [wt1, wt2]

Node

const ta = require('wickra');
const w = new ta.WaveTrend(10, 21, 4);
const o = w.update(101, 99, 100); // null during warmup, else { wt1, wt2 }

Interpretation

WaveTrend is a CCI-like typical-price oscillator with extra EMA smoothing, designed to be read at its extremes:

1. Crossover at extremes. The canonical LazyBear signal is wt1 crossing above wt2 while both are oversold (wt1 < −60), and the mirror in the overbought zone (wt1 > +60). Mid-range crossovers are weak.
2. Divergence. Price making a new high while wt1 does not is a classic momentum-divergence reversal cue.

Common pitfalls

• Trading mid-range crossovers. WaveTrend is intentionally mean-reverting; signals away from the ±60 zones are low-conviction.
• classic() returns a Result. Unlike most classic() constructors in the library, WaveTrend::classic() is fallible — unwrap or ? it.

References

• LazyBear, "Indicator: WaveTrend Oscillator [WT]" — TradingView Pine script, 2014. The construction mirrors the typical-price CCI with an added EMA smoothing on the 0.015-scaled deviation.

See also

• Cci — the typical-price oscillator WaveTrend generalises.
• StochRsi — another bounded mean-reverting oscillator.
• Smi — Blau's doubly-smoothed stochastic momentum.