Here is an interesting paper on using Fractal Market Hypothesis to predict the state of the market. any one want to try to code the indicator for testing?
I jsut got my new VAIO ! such a light in the darkness! yes, I could try to do something. I have a nother stuff, very important, the EA Grid, inf first aim
I have a friend specialized in laptop hardware, and according to him from inside VAIO are very solid machines and with very well manufactured internal components.
The article is nice but still I do not know how to make a mql code for orthogonal linear regression. And I am a bit skeptik (a little bit) about the way you transform the fractal dimension into directionnal indication.
In the provided example you can see very nice trending markets (as usual) and just any indicator is going to perform on them.
Thank you Jaguar, but no one has to feel obliged if the instructions are not clear about how to make it, or he thinks that the approach is not enough perspective.
I think that the article is great however its weakest part is about the code.
I read through the article, and it seems to me that what is called the Fourier dimension is closely related to the Hurst exponent. From what I understand, the core of the indicator revolves around the behaviour of the derivative of this Fourier dimension which is exactly equal to 2 times the derivative of the Hurst exponent. I have provided an indicator that plots the variation of the hurst exponent, and I doubt it provides a very good idea of the direction.
Actually more generally, I believe technical analysis is not meant to identify the direction of the market, this feature seems to me more correlated with news analysis, and would therefore requires something like an AI and a measuring procedure of the news impact. Technical Analysis seems more likely to reveal deep-seated features of the market dynamics (such as resistance and support), and being based on more permanent symmetries, they cannot account for the direction which is a direct translation of the market sentiment, and therefore more dependent on fundamentals.
As for the OLR Algo, it's just a curve fitting algo, it actually replaces the Least Squares method in the Rescaled Range analysis, so, it does not change in any way the spirit of the indicator.
As you know maybe it exist a Matlab coder package that convert .m files to .h ones for C++ or C# (need to check). Does anyone have eventually a such expensive package ?
Comments
Yes,
I jsut got my new VAIO ! such a light in the darkness! yes, I could try to do something. I have a nother stuff, very important, the EA Grid, inf first aim
I have a friend specialized in laptop hardware, and according to him from inside VAIO are very solid machines and with very well manufactured internal components.
The article is nice but still I do not know how to make a mql code for orthogonal linear regression. And I am a bit skeptik (a little bit) about the way you transform the fractal dimension into directionnal indication.
In the provided example you can see very nice trending markets (as usual) and just any indicator is going to perform on them.
Yes, I know it"s difficult, but it's also the purpose of this website to show this kind of algorythm, and ask people like me to code this.
I try my best effort to understand this and code it. Not sure about the result. I will try
Thank you Jaguar, but no one has to feel obliged if the instructions are not clear about how to make it, or he thinks that the approach is not enough perspective.
I think that the article is great however its weakest part is about the code.
Of course, the description is unclear
The algorythm is based on the "Plus least squared regression linear".
If you could get the code in C, I can try more easily to rewrite it in MQ4.
I tried to make something but i failed miserably.
hello all,
i found some lecture notes by the the author.
http://konwersatorium.pw.edu.pl/wyklady/2010_VLZ7_01_wyklad.pdf
More papers can be read from his website here
http://eleceng.dit.ie/blackledge
Great !! The demonstration is powerfull
Hi
Anyone could provide the OLD algorithm ?
rgds
I have no idea. Sorry.
OLR algorithm ?
Sorry no idea how to implement this in mt4, and i do not have any idea where we would find it.
Hi all,
I read through the article, and it seems to me that what is called the Fourier dimension is closely related to the Hurst exponent. From what I understand, the core of the indicator revolves around the behaviour of the derivative of this Fourier dimension which is exactly equal to 2 times the derivative of the Hurst exponent. I have provided an indicator that plots the variation of the hurst exponent, and I doubt it provides a very good idea of the direction.
Actually more generally, I believe technical analysis is not meant to identify the direction of the market, this feature seems to me more correlated with news analysis, and would therefore requires something like an AI and a measuring procedure of the news impact. Technical Analysis seems more likely to reveal deep-seated features of the market dynamics (such as resistance and support), and being based on more permanent symmetries, they cannot account for the direction which is a direct translation of the market sentiment, and therefore more dependent on fundamentals.
Cheers
As for the OLR Algo, it's just a curve fitting algo, it actually replaces the Least Squares method in the Rescaled Range analysis, so, it does not change in any way the spirit of the indicator.
Hello Jean-Philippe,
can you upload the indicator which you have written?
thank you.
-guandi
for FGDI
http://codebase.mql4.com/5525
For Fractals _bands, check the web
for FGDI
http://codebase.mql4.com/5525
For Fractals _bands, check the web
Thank you Jean-Philippe for the answer. Guandi the link is here, the indicator is named Hurst difference.
Bonjour à tous et bravo pour votre travail.
I'm also interested int the q-algorithm as I read all the blackledge papers.
The ORL is a Deming regression considering delta=1. The estimators can be found here http://en.wikipedia.org/wiki/Deming_regression (another way is to calculate via matrix).
As I'm not a great coder, I think after calculate the moment qj=1+2Hj to calculate new qj=qi with the previous estimators and the real xi & yi.
Then applying the different moving averages.
If it could help some, i would be interested in the mql4 code.
Yep, I am too interested in coding in MQ4L,
I need the Code in C langage of this algorithm for this purpose
As you know maybe it exist a Matlab coder package that convert .m files to .h ones for C++ or C# (need to check). Does anyone have eventually a such expensive package ?