Log in

A Holy Grail? No just a low fractal dimension

By JohnLast 3237 days ago Comments (10)

A Holy Grail? No just a low fractal dimension

 

image
От 22 септември 2011

Comments

  • sachinarora 3237 days ago

    hello sir

    please specify the name of indicator attached in bars and parameters.

    thanks

  • JohnLast 3236 days ago

    This is the Brain Trend SSA en pointed. I think it is the default parameters. I just loaded the templetate and that was the result. However Brain Trend SSA ep has one issue. It does not repaint but not always is what you get on the screen is what you get that is why the change of parameters would  be misleading.

    This indicator is in the TSD where I uploaded it, but I will upload it here too.

    Look at here this the brain trend double smoothed the results are not much different. What REALLY matters is the fractal dimension. Look at the iVAR its readings are mostly below 0.5. That is why Brain Trend works. That is the oldest trick of indicators developpment to show an easy time series (low fractal dimensional time series) series and apply the indicator on it (the second trick you all know it is to optimize the parameters).

    That is why indicators and systems fail and do start to work again, but everything depend on the underlying market structure. Is it within the parameters of you system or not?

      

     

    image
    От brain trend
  • tovim 3234 days ago

    John ,

    There is an interesting idea here  in the link:  http://hanwangquant.blogspot.com/2011/07/fractal-volatility-through-variation.html

    It is  fractal volatility through variation index

    fractal volatilty indicator was already coded in mql4  http://www.expertadvisordownloads.com/category/indicators/sub-window-indicators/bar-histogramm-indicators-sub-window-indicators/page/3/

     

  • JohnLast 3234 days ago

    Hi,

     

    I will check but at the first look the ivolty is a different thing  from variation index ivar

    In fact thanks at last it is possible to share this article in English, the original is in Russian.

     

  • tovim 3234 days ago

    Hi John

    Fractal volatility indicator in mql4 in the link I have given, based on box counting(fractal geometry),the blog owner brought new idea by  the new fractal volatility indicator based on fractal dimension index(in this case I-VAR indicator) instead of box counting http://classes.yale.edu/fractals/fracanddim/boxdim/BoxDim.html

  • JohnLast 3234 days ago

    Do you have any idea how it works because from the first look I did not find any function MathLog (The MathLog function returns the natural logarithm of x if successful.) as I expected. Or maybe I am wrong.

  • jaguar1637 3234 days ago

    Hi, the MathLog returns the natural logarithm like this :

    z = MathLog;            => only if S is > 0

    it works like   y = MathSqrt(F);

    Regarding the fractal volatility in the previous post, the results in trading are quite average.

     

  • francisfinley 3232 days ago

    as you know i spent a long time working on brain trend & ssa - no joy on forex.

    maybe i will do better with it on DAX

  • JohnLast 3228 days ago

    Hi, francisfinley are you selling it or you are going to share?

     

  • JohnLast 3189 days ago
    The fractal dimension indicators give a probabilistic measure if your directional indicators are going to work. For example if the Hurst Exponent is bigger than 0.5 there is a bigger probability that the next movement is going to be in the same direction as the previous.
    And vice versa if the Hurst exponent is below 0.5 the next movement is more probable to reverse.
    So this is a basic clue about predictability. The iVAR has a reverse scale than the Hurst exponent, when iVAR below 0.5 the Hurst exponent is above 0.5 and vice versa.
    And you could guess that a procedure in order to detect (evaluate) a fast change in the Hurst exponent from a small dataset is critical. 
    Of course we did made experts and included that as a procedure for decision making in order to take a decision to enter or not in a positions. The results were interesting. In fact this procedure did not improve the profitability however this procedure made the equity line smoother. 
    The second basic approach was to use this together with the model of court faith combining fractal dimension with human pattern recognition and volatility. 
    What is interesting. The volatility has its daily own cycles. The fractal dimension state has its own cycles too that do not coincide with the volatility cycles. 
    Look ath the screenshot at the beginning of the page again.
    This is the S&P. The idea is to use for prediction the horizon where the hurst exponent is the highest. Here I explain the I had wonderful signals not because I have a holy grail but because the market state was with particularly low fractal dimension (large hurst exponent) and that made my impulse following algorithm to perform nicely. 
    And why this happen. The hypothesis is that we can analyse the market as a problem to which many humans and machines are looking to find the right answer. The problem can be explained by multidimensional phase space of the possible solutions. 
    Sometimes the problem can be solved simultaneously by many machines and humans. And they start driving the prices in one particular direction. (this is purely theoretical in practice this happen very often when a lot of humans have their stops activated at particular places and they start to be runned one after the other).
    This fractal state is characterized by black noise reversals, sharp V bottom and V tops?