Our time frequency analysis is a bespoke technique which utilises techniques from several scientific fields. In this article we explain some of the essential aspects of these charts for subscribers
This is very interesting. I use the terms “wavelets” for a different predictive method, which I have used over the years in my model.
I was screening Google for older entries using “wavelets” and came across yours, all the while as I was just waiting to receive a book on Hurst.
I love the coincidence, and I was so intrigued, I decided to join. Very intrigued still. I’ll continue to follow.
My handle on Twitter is 4xForecaster and the model I use is called Constant Rescaling Of Wavelets (C.R.O.W.) - Feel free to comment there, if ever interested as well.
Again, thank you and I’ll continue to follow through a well-worth paid membership here.
Thanks for your kind words and information David. In the context of the analysis on Sigma-L wavelets are 'morlet wavelets' - essentially a sine wave mixed with a Gaussian as the 'mother' wavelet. This retains both temporal and frequency information from the signal, the mixing of the two defined by the parameters of the Gaussian and the frequency/scale of the wavelet itself. Other wavelet types are available of course but the frequency response of the Morlet is a pure Gaussian itself, which has distinct advantages (although is sometimes a little too smooth!).
Interesting. The wavelets that I reference in my analysis are up/down scales of same-selves iterations seen across a same interval.
To put it simply, wave patterns a reiterated India’s a specific morphology, say from a 4-hour chart, whose “tail” is reiterated in a 1-hour, whose tail is also reiterated in a 15-min, such that these iterations remain visibly present in up/down temporal scales (timeframes) at a multiple of 4 more or less:
M15 x 4 = H1
H1 x 4 = H4
H4 x 4 ~ Daily
Daily x 4 ~ W
W x 4 = M
What I find most curious is that we’re arriving at similar reversals.
Although I’ve studied these price reversals for years, it’s only a few years back when I started to focus on time-reversals.
Of late, I delved into Hurst analyses; that’s how I came across your analyses.
Fascinating and very intriguing. Been a while since I found something keeping me up late at night.
Thanks David! Very detailed intro! I have a small question if you have time to answer: generally speaking, should we use return or price as input to the time frequency analysis? Some claim using return gives better frequency information and some claim using price retains more memory, thereby gaining more predictability.
Very good question. I have looked into this myself and intend to build in returns to the system over the UK winter or sometime next year. Whether there is periodic activity in returns remains to be seen and I will be guided by the objective evidence. If there is evidence of some kind of momentum based periodic activity that could be very useful when used in conjunction with time based periodicity, which is evident in the current time frequency analysis on site.
Thank you David,
This is very interesting. I use the terms “wavelets” for a different predictive method, which I have used over the years in my model.
I was screening Google for older entries using “wavelets” and came across yours, all the while as I was just waiting to receive a book on Hurst.
I love the coincidence, and I was so intrigued, I decided to join. Very intrigued still. I’ll continue to follow.
My handle on Twitter is 4xForecaster and the model I use is called Constant Rescaling Of Wavelets (C.R.O.W.) - Feel free to comment there, if ever interested as well.
Again, thank you and I’ll continue to follow through a well-worth paid membership here.
David Alcindor
Wyoming, USA
Thanks for your kind words and information David. In the context of the analysis on Sigma-L wavelets are 'morlet wavelets' - essentially a sine wave mixed with a Gaussian as the 'mother' wavelet. This retains both temporal and frequency information from the signal, the mixing of the two defined by the parameters of the Gaussian and the frequency/scale of the wavelet itself. Other wavelet types are available of course but the frequency response of the Morlet is a pure Gaussian itself, which has distinct advantages (although is sometimes a little too smooth!).
Interesting. The wavelets that I reference in my analysis are up/down scales of same-selves iterations seen across a same interval.
To put it simply, wave patterns a reiterated India’s a specific morphology, say from a 4-hour chart, whose “tail” is reiterated in a 1-hour, whose tail is also reiterated in a 15-min, such that these iterations remain visibly present in up/down temporal scales (timeframes) at a multiple of 4 more or less:
M15 x 4 = H1
H1 x 4 = H4
H4 x 4 ~ Daily
Daily x 4 ~ W
W x 4 = M
What I find most curious is that we’re arriving at similar reversals.
Although I’ve studied these price reversals for years, it’s only a few years back when I started to focus on time-reversals.
Of late, I delved into Hurst analyses; that’s how I came across your analyses.
Fascinating and very intriguing. Been a while since I found something keeping me up late at night.
Thank you.
Thanks David! Very detailed intro! I have a small question if you have time to answer: generally speaking, should we use return or price as input to the time frequency analysis? Some claim using return gives better frequency information and some claim using price retains more memory, thereby gaining more predictability.
Very good question. I have looked into this myself and intend to build in returns to the system over the UK winter or sometime next year. Whether there is periodic activity in returns remains to be seen and I will be guided by the objective evidence. If there is evidence of some kind of momentum based periodic activity that could be very useful when used in conjunction with time based periodicity, which is evident in the current time frequency analysis on site.