Log in

SSA EA example

SSA EA example
  • Public
By JohnLast 2325 days ago Comments (4)

This is an example  how to call directly the SSA dll for practical use in EA. This is not a working trading EA, just an example how to implement the idea.

Comments

  • jaguar1637 2320 days ago

    Hi

    Now, we got this pratical point for removing false entries. We used iVAR in TF 15MN

    but jc shows now to us, that iVAR should be used in TF 1mnute (1MN), he told me that removes twice the bad entries with this system. the security level taken is something around .055

  • JohnLast 2316 days ago

    I think I remain misunderstood on what I am trying to focus the attention. 

    I know it is damn complicated but there is hiding something important. 

    How to overcome the data mining bias and build a robust system? That is the question.

    The back  - testing and forward testing methods are obsolete by now.

  • jaguar1637 2316 days ago

    Hi John

    Nope , everything is on the table

    We got iVAR to remove the false entries

    Hodrick Prescott and SSA or fetch a solid time-series.

    A simple MA crossing on SSA w/ diferent lag (8 vs 25) works

    The issue remains in the fact, I observed 3 types of market

    - a flat market, a trending market and a high volatility market, that implies 3 different strategies and 3 # logics

     

  • JohnLast 2315 days ago

    I was thinking again about the market state identification. And perhaps the simplification by making a clear cut classification of the market state is wrong.

    As we remember we have according to curtis faith 4 clear market states with combinations.

    Here is the page describing that stuff.

    http://beathespread.com/pages/view/438/curtis-faith-and-the-market-state-analysis

    So if we have a four market states let say A,B,C,D types.

    A market state could be described for example as 25 % A, 30 % B, 20 % C and 30 % D.

    So in practice in order that our hypothetic system to work we may need two conditions:

    At least 20 % of market state of C (for example trending and quiet)

    And no more than 25 % of A (stable and quiet range).

    I think this describes much better the reality of the rise and fall of trading systems.