ultimate short friday?, page-91

Checked out the frequency fits using a general least squares prog, think it just uses steepest descent method.

Fitted Log(SPX) since 1950->Present with linear function, extracted residual, then fitted a single frequency, get ~37yrs.


Extracted info from this again to get higher frequency modes with average wavelength of 3.9yrs:



Lines up well except for couple cases.

Focused in on recent few years, gets bit messy there, so just did smoothing function, and looked at difference to get higher frequencies.


Gets interesting, as can do aprox fit with a Gaussian modulated sinusoidal function...get wavelength of...3.96 months, give or take.
Common signal, even used in medical diagnostic equipment: Ultrasound.



Three fractal levels, decades (3.7 +-), years (3.9, Standard Deviation~0.2) and months (3.96+-), lol.

Calling it the 3.9 LunchHour Model, is there a large ~4h signal in markets???.