← All indicators

zeonta.reflex_trendflex() — Ehlers’ zero-lag pair: deviation from a fitted line (cycle) vs. current value.

What it measures

Both start from the same super_smoother pass at half length, then average that filtered line’s own deviation from a reference over the full window — Reflex measures deviation from a straight line drawn across the window (stripping trend, isolating cycle swings), Trendflex measures deviation from the filtered line’s current value (keeping trend in).

Formula

Filt = SuperSmoother(Close, length/2); Reflex = mean(Filt+k*Slope-Filt[-k]) / sqrt(MS); Trendflex = mean(Filt-Filt[-k]) / sqrt(MS)

Parameters

Required inputs: close

Parameter	Default
length	20

Returns

Column
REFLEX_20
TRENDFLEX_20

Usage

Examples run against the 300-bar OHLCV fixture in tests/data/ohlcv.csv, loaded as df. The output shown is the real output.

import pandas as pd
import zeonta

df = pd.read_csv('tests/data/ohlcv.csv', parse_dates=['date']).set_index('date')

zeonta.reflex_trendflex(df['close']).tail(3)

            REFLEX_20  TRENDFLEX_20
date                               
2024-10-25  -0.119571     -0.845713
2024-10-26  -0.412141     -0.866909
2024-10-27  -0.910793     -1.203085

Accessor form: df.zta.reflex_trendflex(...)

How to read it

Both self-normalize against their own recent mean square, the same “divide by local RMS” idea even_better_sinewave uses, so their scale stays comparable across different volatility regimes — read zero-line crossings and extremes the same way you would any zero-lag oscillator.

Pitfalls

Reflex and Trendflex answer different questions from the same input — Reflex isolates the cycle, Trendflex keeps the trend — so reading one where you meant the other gives a misleading signal even though both are always well-defined together.

Reference

Formula source: https://www.prorealcode.com/prorealtime-indicators/reflex-and-trendflex-indicators-john-f-ehlers/