FRAKTAL·ICIOUS
Live reading
Model signal · updated 2026-08-25

Weekly Gaussian channel

A four-pole filter on weekly bars, read as position plus slope

On 2026-08-25 the Fraktalicious cycle model scored the Gaussian channel at −0.400 — cautionary — and gave it an effective weight of 0.020. That is a contribution of −0.0080 to the day's raw composite of −0.0060, the 7th-largest of the 15 signals carrying weight today.

Signal score
−0.400
on the model's −1 to +1 scale · higher than 17.2% of its own graded history
Effective weight
0.020
regime-scaled · 0.031 on average since 2012-10-27
Contribution today
−0.0080
rank 7 of 15 active signals

What the Gaussian channel is

A Gaussian channel is a smoothed trend line with a volatility envelope around it. This one runs a four-pole Gaussian filter, with a sampling period of 144, over the average of each week's high, low and close; the envelope is that filtered mean plus and minus 1.414 times a filtered measure of weekly true range.

The filter is heavier than a moving average and lags accordingly. That is the trade being made: it will not tell you about last week, but the line it draws does not wobble, so the direction of the line is meaningful on its own.

How this model uses it

Only two facts are extracted, and the score is the eight-way combination of them: where the weekly close sits relative to the channel (above the upper band, above the mean, below the mean, below the lower band), and whether the filtered mean rose week-over-week.

The pairing is what makes it a signal rather than a trend follower. Above the upper band with a rising mean scores +0.5; the same position with a falling mean scores −0.3. Below the lower band with a rising mean is a constructive +0.3, but below the lower band with a falling mean is −0.7 — the most cautionary reading it produces. The model treats a broken trend as a reason for caution rather than as a discount.

It belongs to the timing family and is damped where a smoothed trend line is least useful: halved in distribution and in a bear market, cut to under a third in accumulation, left at nominal in a bull or a parabola.

Where today sits in its own history

Every figure in this table is measured over the model's full graded replay — 5,051 days from 2012-10-27 to 2026-08-25 — and re-measured on every daily refresh.

MeasureValue
Score today−0.400
Percentile of that score in its own history17.2%
Most constructive reading on record+0.500 · 2012-11-11
Most cautionary reading on record−0.700 · 2014-12-21
Days it carried weight5,051 of 5,051
Share of those days its score changed1.1%
Mean effective weight, over the days it was active0.0308
Share of all weight the model assigned, averaged over every day5.4%

What this does not tell you

The composite this metric feeds cut maximum drawdown in 18 of 18 walk-forward folds (the year and cycle folds tile the same span, so they are not independent of each other; this is a consistency check, not proof of timing skill). It beat buy-and-hold on return in 11 of 18 of those folds, which is close to a coin flip. Drawdown reduction is the claim this project stands behind; return is not.
Figures on this page are regenerated from the graded replay on every daily data refresh. If you are citing one, cite it with its date — it will have moved by the time anyone reads you.
Fraktalicious Research · Metric index · Call ledger · Briefing archive · Free widget & API · Privacy
Published research on a fixed schedule, identical for every reader. Nothing here is financial advice, nothing here is personalised, and nothing here is a price prediction.