Build an adaptive oscillator (macci) with good measurement of dominant market cycle

[Gob00st]

Hi Guys,
It's been quiet in TT for quite a while, I am wondering if people here have got much advanced strategy than multiple time frame analysis with static macci (5,6)?

After I spent some time study John Ehlers's public indicators in his MESA website, I started wondering if an adaptive AMacci(x,y) would perform better than the static macci(5,6) ? So I take one step and tried to use the half of the dominant cycle measured by Fourier Transfer as input x and half of x as y ( not sure if this is right).

The result sometime seems encourage but sometime is pretty useless.
I am not sure if this is due to the dominant cycle measured by Fourier Transfer is not so stable/ accurate? Of course I have been only tried to forward test when market is in cycle mode, I have been trying to identify that with the help of instantaneous trendline & sinewave.

I am wondering if there is anyone in TT has been trying to do similar thing in order to build an adaptive macci and willing to share some thoughts.

Cheers
Albert

 

[dcraig1]

Hi Guys,
The result sometime seems encourage but sometime is pretty useless.
I am not sure if this is due to the dominant cycle measured by Fourier Transfer is not so stable/ accurate?

My guess is that is case. (which is only a guess). Sorry can't really help much. If it's any consolation, this is a pretty tough nut to crack.

 

[Gob00st]

What do you mean if its any consolidation ?

My guess is that is case. (which is only a guess). Sorry can't really help much. If it's any consolation, this is a pretty tough nut to crack.

 

[dcraig1]

What do you mean if its any consolidation ?

Oops. Sorry ... I meant "consolation". I've coded up some of the Ehlers freebees in Java and I must say I don't think I got anywhere with them. Though I didn't try to make an adaptive cycle indicator.

I did try some of the cycle strength and period stuff integrated into my stock screener, but couldn't find any edge really. I like to try things in the screener 'cause it allows me to look at thousands of instruments.

I think Jurik talks about adaptive cycle stuff somewhere on his web site using his proprietary (for money) indicators. I think he uses his "compound fractal dimension" indicator to determine period. I don't know what the latter is, but guess it might be sum of estimates of fractal dimension over multiple periods. Fractal dimension is not to hard to calculate. It might be worth looking at.

 

[Gob00st]

dcraig,
Thanks for your consolation.
For Ehlers freebees, I found IFisher could be quite useful in higher timeframe in my limited study of them.
I did a quick google for Jurik, the only freebee I found is their special report, but I didn't find anything related fractal dimension.

It seems you have done quite a lot of work for stock screening, may I ask what you use for market direction call ? I think in top down approach, market direction is still the first step?

Regards,
Albert

Oops. Sorry ... I meant "consolation". I've coded up some of the Ehlers freebees in Java and I must say I don't think I got anywhere with them. Though I didn't try to make an adaptive cycle indicator.

I did try some of the cycle strength and period stuff integrated into my stock screener, but couldn't find any edge really. I like to try things in the screener 'cause it allows me to look at thousands of instruments.

I think Jurik talks about adaptive cycle stuff somewhere on his web site using his proprietary (for money) indicators. I think he uses his "compound fractal dimension" indicator to determine period. I don't know what the latter is, but guess it might be sum of estimates of fractal dimension over multiple periods. Fractal dimension is not to hard to calculate. It might be worth looking at.