Gaussian Filter

Gaussian Filter smooths price by running it through the same simple smoothing stage up to four times in a row, so the cascade's combined response approaches the Gaussian bell and the line comes out round and steady. This build follows John Ehlers' published recursive recipe for the Gaussian filter: the per-stage coefficient is derived from the period and pole count, and the selected pole's output is the plotted line. The result reads as a trend baseline — slope carries the regime, and price's side of the line frames the structure around it.

How to Trade the Gaussian Filter?

• Slope color: with Slope Trend Coloring on, the line is bullish while rising and bearish while falling — slope turns are alertable in both directions and make clean regime markers.
• Price crossings: the standard moving-average crossover grammar on a smoother line, with fewer whipsaw crosses than a raw average; optional triangles mark them, and both directions are alertable.
• Fill direction: the gradient between price and filter shows which side price is riding — persistent one-sided fills characterize trends, alternating fills congestion.

Gaussian Filter Settings

• Source (default close): the series being smoothed; feeding another indicator's output through it covers the pre-smoother use.
• Period (default 14): the cutoff the per-stage coefficient is derived from — longer gives a rounder, slower line, shorter tracks price but passes more noise.
• Poles (default 4): identical stages in series. Each extra pole adds smoothness and lag at the same period and pushes the response closer to the bell; one pole behaves much like an EMA, and four is the cap in Ehlers' original.
• Style: Slope Trend Coloring (on), Price / Filter Gradient Fill (on), Price Crossing Markers (off), plus bullish, bearish and neutral colors.

Frequently Asked Questions

Gaussian Filter or the SuperSmoother?

Both are Ehlers designs chasing smoothness without runaway lag. The Ehlers SuperSmoother is a two-pole design built to reject fast noise while giving up as little speed as possible, while the Gaussian cascade accepts slightly more lag for a bell-shaped, ripple-free response — and its pole count gives an explicit smoothness dial.

How should I choose the pole count?

Treat it as a smoothness dial at fixed period: one pole for responsiveness, four for the roundest line. If you find yourself shortening the period to offset added lag, try fewer poles instead — the two settings trade off differently.

Does the filter repaint?

No. It is a causal recursion over current and past values only — once a bar closes its value is fixed, and alerts evaluate on the same series.