Reading Millard, I’ve finished ‘Channels and Cycles’ and now reading ‘Future Trends from Past Cycles’.
I think Millard has a praiseworthy approach to dealing with Hurst’s basic cyclical ideas, abandoning (or at least diminishing) the rigid formal structure and preferring the use of bands which are both more flexible and less work than manual phasing. I think it’s still worthy of debate whether a band can actually predict future price direction, but here is an idea I haven’t encountered in Hurst or Millard (yet) and wondered if it might be useful to y’all.
The simple premise is that only channels/bands that are horizontal can give us reliable info.
Sideways price segments correspond to ranges (Wyckoff distributions) and can be found with a standard deviation channel - just adjust either side of it to enclose data such that it becomes as horizontal as possible - there’s your range.
So, I’d hoped to find that cycles within a range are more stable than those found in entire price histories. Certainly I’ve seen that cycle periods can yawn wide on either side of a strong trend. And it would seem logical that during a ‘resting’ period price would more closely follow cyclical principles since it’s (likely) un-motivated price action.
Using (Mandelbrot’s) principle of multifractality you only need zoom out once your channel/bands start sloping and you can find another (larger) range that will give you a reasonable estimate for the next reversal and a stable cycle period. Furthermore only cycles of the size found within that range would be valid for future estimates. This could provide a theoretical basis for the appearance/disappearance/re-emergence of dominant cycles. After a strong trend that obliterates the dominant cycle, the range re-asserts itself along with the prior cycle. If you wanted to trade cyclically on a smaller TF you’d need to find a smaller range.
Another idea, maybe easier to prove is that only band-to-band touches of horizontal bands will make for valid (reasonably regular) average measurements of amplitude and duration. So if you want to gauge the duration of a current trend you need to examine it from the stability of a stable section of a higher degree cycle.
Even better, if you find that the cyclicality for your selected period suddenly is violated that would mean advance warning that price has entered a trend.