EA needs to be refined - page 4

New comment
 
Mischek:
No, in 2b.
I'm talking to Alexander. You shouldn't have read it at all. DANGER

FIRE-DANGEROUS... or just?
 
vladimir.kuc:

FLAMMABLE... or simply ?

Do you know how to intrigue a freeloader for a week?
 
Mischek:

Do you know how to intrigue a freeloader for a week?

Better a girl.

She'll do more good.

 
Here you (and us) guys have been subtly but popularly explained that a programmer is worse than a woman.
 
vladimir.kuc:

In 1) parameters are optimised block by block, i.e. step by step in 4 steps, with no loss of quality,

2) it is possible to optimize by the opening prices, this is how the Expert Advisor is built, which gives a breakthrough in speed.


There is a logic of action, temporal regularities, etc.

The question is not about the methodology of optimisation, but about the number of EA parameters. I understand that it is hard to hear unflattering comments about a product which took 2 years to develop. But still: The opinion of most developers comes down to the fact that an ideal TS should not have any parameters at all. It should be adaptable to the market, i.e. the values of the parameters should not be set by external variables, but calculated inside the TS on the basis of some data. Each external (manually set) parameter deprives the TS of flexibility and at the same time allows to fit the results on a certain area of optimization. Accordingly, the more external parameters there are, the more accurately the balance/equity curve can be fitted to the history. And the better it is fitted in the optimization area, the less chance of success on the forward side.
 
vladimir.kuc:

Better a girl.

She's better for you.


Then you don't know. I'll tell you in a week.
 
 
goldtrader:
It's not a question of optimization methodology, but of the number of EA parameters. I understand that it is hard to hear unflattering comments about a product development of which took 2 years. But still: The opinion of most of the developers comes down to the fact that an ideal TS should not have any parameters at all. It should be adaptable to the market, i.e. the values of the parameters should not be set by external variables, but calculated inside the TS on the basis of some data. Each external (manually set) parameter deprives the TS of flexibility and at the same time allows to fit the results on a certain area of optimization. Accordingly, the more external parameters there are, the more accurately the balance/equity curve can be fitted to the history. And the better it is fitted in the optimization area, the less chance of success on the forward side.

who says these parameters are static?


Each clone makes its own decision, and based on its parameters and the decisions of the 2nd ... 3rd ... 4th clone makes its own and vice versa ... - it is a linked non static network ...


floating parameters - at basic indicator ... at filters,


floating TP/SL


floating time SLs ...

floating time TPs ...


everything just adapts to the market ...

 
Swetten:


Well needless to say, there are no working solutions...

 
Swetten:


Where are the drawdowns?
12345678
New comment