Skip to main content
Track Adaptive Filter
4 / 9
Library

2000issue C111-3

When charting code breaks the adaptive trendline

Readers who coded a published instantaneous trendline and adaptive oscillator reported dual-plot artifacts and traces that did not match across commercial platforms. Editorial view: treat the specification as incomplete until two independent engines, given the same series and lookback, draw matching traces.

  • Treat a published instantaneous trendline or adaptive oscillator as incomplete until two independent engines, given the same series and lookback, draw matching traces.
  • A dual-plot artifact, including a price overlay that cannot be restyled on its own, is an implementation-fidelity failure rather than a market signal.
  • Cross-platform-reproduction isolates a code mismatch from the underlying trendline and adaptive-filter methods.
  • Archive correspondence treated the studies as usable once implemented correctly and framed the barrier as formula presentation.
Entries in this reading2 entries

Dual-plot artifacts in one package

A reader who applied a published adaptive-trend and oscillator specification in one retail charting package obtained two identically titled plots that could not be restyled independently. The same reader found that the companion instantaneous-trendline formula also drew two traces, one of which appeared to overlay the price series.

Those extra lines are a dual-plot artifact: an unintended second line in a study pane, often a price overlay or paired component that cannot be styled or read separately. The letters present the extra traces as a plotting problem.

A trendline here is a chart-scale line taken from price structure, including an instantaneous construction meant to track the current trend rather than a hand-drawn support or resistance segment. An adaptive-filter is a quantitative smoother or oscillator whose coefficients or lookback change with a measured cycle or sampling interval, meant to be checked against an explicit baseline.

Traces that did not match

A second reader compared the published instantaneous-trendline and sine-wave companion across two commercial platforms and reported that the outputs did not match. That second reader argued that a code mismatch could lead practitioners to discard the underlying trendline and filter methods rather than isolate an implementation error.

Cross-platform-reproduction is the check of whether two charting engines, given the same specification and price series, produce matching trendline and oscillator traces. Implementation-fidelity is agreement between a published formula and the series a given software package actually plots.

Referral rather than rejection

The letters editor declined to debug the formulas in-column and referred both implementation complaints to the charting-software vendor. The correspondence treated the adaptive-trend and instantaneous-trendline studies as usable once implemented correctly, and framed the barrier as formula presentation rather than a rejection of the methods themselves.

Editorial. The archive facts describe that historical workflow of letters and vendor referral. TradersWeek reads the same record as a specification that remains incomplete until two independent engines, given the same series and lookback, draw matching traces.

Educational research material, not investment advice. Historical source context does not establish present-day performance.
4 of 9 in the Adaptive Filter track
20011-6 pp.Next on Adaptive FilterNonlinear Ehlers filter construction from momentum and distance weightsThe nonlinear construction forms each coefficient from a chosen statistic of the price series, multiplies those coefficients by the corresponding prices, and outputs the ratio of that numerator to the coefficient sum.
All readings on this track · 9 readings
  1. 1982Constructing moving averages with weights, alignment, and adaptive lookbacks
  2. 1990Constructing adaptive filters from equal-weight averages
  3. 2000Cycle-scaled lookbacks for a channel-breakout rule
  4. 2000When charting code breaks the adaptive trendline
  5. 2001Nonlinear Ehlers filter construction from momentum and distance weights
  6. 2001How a fixed smoother becomes a living parameter surface
  7. 2005A three-part workshop for adaptive-filter length search
  8. 2010Rebuilding an exponential average as a bounded error-correcting filter
  9. 2018Assembling a recursive median oscillator as a two-stage adaptive filter
All 9 readings tagged Adaptive Filter
Also on Adaptive Filter5 readings