Log in

Spinal implant with iVAR kernel

Spinal implant with iVAR kernel
By JohnLast 3045 days ago Comments (5)

This is a test version of the Spinal implant with kernel adaptation of iVAR combination.

The idea is that we can use a combination of iVARs with different period of estimating the Hurst exponent. 

The idea is not very theoretically sound but, nevertheless I made a trial. 

So this is the EA with the set file for 15 m. the same set file surpisingly gives an identical equity without being optimized for the time frame.

So this is you can test it.


От EA tests


  • JohnLast 3030 days ago

    Just to mentit mention, I did not make attention. But I will let it as it is. Here I use with the same set the same EA but with money management module that is included. 

    So it makes the same trades but the results are better (the equity curve is different) because of the applied money management. (The money management  is described here). In a nutshell the lot size is decreased after a loosing trade.

  • jaguar1637 3030 days ago

    Well, I thought about a iVAR Perceptron w/ a smoothing tool. But, I think now, the smooth function should not be used.

    Not sure a perceptron is required. nor a sample function for iVAR perceptron (not gaussian)

    I think the iVAR function proposed by John is enough.

    The entry/exit confirmation signals can be provided by other kind of perceptrons

  • JohnLast 3030 days ago

    I see, I had this concern but as you can see in the code:

     double w1 =  MathExp(gamma*(-x1*x1-y1*y1))+1;

    As the iVAR variates between 0 and 1 it makes no sense at all to multiply it by a weight itself between 0 and 1.

    That is why I add a 1 for the weight, this one will make the weight between 1 and 2.

    So here for example we can adjust the iVAR in the averiging. For example we multiply it by 1,12 or 1,05.

    And then we make an averaging:

     return((p1*w1*w2 + p2*w1*w2 + p3*w1*w2)/3);

    Anyway this solves the problem, what is the most usefull averaged Hurst estimation ;).


  • jaguar1637 3030 days ago

    ok I will post a new release of PerceptronIndicator_iVAR

  • jaguar1637 2961 days ago

    Because there is no correlation betweeb iVAR and market price.

    We should try HP_oscillator instead.