Signs of a REAL system - page 3

New comment
 
ForexTools писал(а) >>

signs of fitting:

    ......
    • results on another TF differ strongly for the worse relative to the baseline,
    • results on other similar (for example, GBPUSD after EURUSD) trade instrument of the TF differ strongly to the worse in comparison with basic ones,
    • the balance curve outside (both left and right) the optimisation area is very different from the view in the optimisation area,
    • presence in the Expert Advisor of a large number (more than 2-3) of optimizable parameters.
     
    RomanS >> :
    An ideal system in which only the lot size and the risk criterion are set in external parameters...

    the lot size is determined by the risk criterion ;)

    the lot size should be calculated under this very criterion and the size of the stop required by the system(Acceptable Risk %)

     
    goldtrader >> :
    • the results on another TF are very much worse in comparison with the basic ones,
    • The results on another similar (e.g. GBPUSD after EURUSD) trading instrument TF are very different for the worse in relation to the baseline,

    It should be noted that the parameters are relative. With absolute parameters, any Expert Advisor cannot be ported to another TF and instrument.

    Not too much? Rather, this is not an evidence of an EA inoperability but rather an irrational (illiterate) approach to optimization.

     
    Svinozavr >> :

    • The presence in an Expert Advisor of a large number (more than 2-3) of optimizable parameters.
    Isn't it too much? This is rather an evidence of an irrational (illiterate) approach to optimization rather than the EA's inefficiency.

    goldtrader wrote >>.

    You're probably right, it's more about the specifics of the EA. But the kind of EA that's SUPPORTABLE.

    If the Expert Advisor has many optimisable parameters, it is easier to adjust.

    At least, I think that this condition should be present in one form or another in order to warn the inexperienced user about the danger of adjustment.

    You can edit it, if it hurts your ears. :)

     
    goldtrader >>:
    • результаты на другом ТФ сильно отличаются в худшую сторону относительно базовых,
    • результаты на другом похожем (например GBPUSD после EURUSD) торговом инструменте ТФ сильно отличаются в худшую сторону относительно базовых,

    What if the difference is for the better? What would that tell you?

     
    joo >> :

    And if the difference is for the better? What would that tell you?

    It's a strong indication that the TC is quite reliable and there's no smell of tinkering.

    But this is not yet a sufficient condition to conclude that there is no fit.

     
    joo >> :

    And if the difference will be for the better? What will it tell?

    ))) In that case...

    In principle, it makes no difference. The essence of the message here is a significant difference in the result. Consequently the TC is extremely sensitive - bad.

     
    Svinozavr >> :

    In principle, it makes no difference. The essence of the message here is a significant difference in outcome. Consequently, the TC is extremely sensitive - bad.

    that's what I was trying to point out

     
    niko1312 >> :

    I would add in "Signs of fitting": difference in results at different intervals of history (for the same DC).

    Maybe my TS is seasonal and I optimize it for last year,

    summer 2008 optimization summer 2009 forward in other seasons this TS won't work so it's a fit?

    >> Maybe my TS takes the seasonality, I optimize it for the last year, at other times another TS works, you do not wear the same T-shirt the whole year round or you do?

     
    Urain >> :

    Maybe my TS is seasonal and I optimize it for last year,

    if I optimise it forwards in summer 2008 or in summer 2009 this EA will not work in other seasons, so why is it a fitting?

    Nothing - the condition on seasonality is a rule of your TS. That is, it is part of the Expert Advisor.

    12345678910...25
    New comment