More strategies? No problem! - page 8

 
danja писал(а) >>
I wish we could use such an algorithm that SSB would look for such a strategy.

I think it's easy to organise.


By the way, I decided now to check how strategies generated on eurusd work on usdchf.

This is what I got:


And the 5th strategy turned out to be the most stable.

 
voltair >> :

I think it's easy to organise.


By the way, I decided now to check how strategies generated on eurusd work on usdchf.

This is what I got:


And the 5th strategy has appeared to be the most stable.

Initial balance? ...... % drawdown is embarrassing

And about easy to organise - very interesting )

 
danja >> :
>> rider. And I haven't seen anyone write that the strategy should show profit on all instruments. I wrote about this idea, but I wanted to do it on several pairs of instruments, of course, if you're talking about my post.

not specifically about yours..... can't remember who or where anymore (the branch turned into a whole tree), but it sounded - it sure did......

And the fact that on several pairs, and even in different TFs - that's absolutely right: a kind of hedging turns out.....

 
rider писал(а) >>

initial balance? ...... % drawdown is confusing

And about easy to organise - very interesting )

Initial balance is $100k, so the drawdown is not more than 25% on the main balance and ~10% on the test balance.


>>Organising it is definitely possible, but I'm not up to it right now.

 
HIDDEN >> :

Not the first day of testing EAs, this generator ran in the tester with a SET file, hung for probably 3 hours, not a single result. the bar does not even move and actually did not show what number of variants of the search.

Try some more, maybe I have a glitch with the terminal. Although other EAs are tested, it means it works.

No. You have to be more patient :).

The whole point is that testing starts at -1 in all conditions.


The maximum allowable number of atomic signals is 5 by default.

The first strategy that will work is 000000000RRRR or 11111111100000 in ternary.

But you don't have to wait for the end, because the last acceptable combination is 2222211111111111. (sorry 1111111111122222)


The catch is that if you don't use a limit on the number of atomic signals in a condition, it will take you many times longer to test.

____________________________

Next. I will ask to create another thread for clarification of generator ratings. And you can in a branch of Mr Reshetov, I hope he will be pleased.

____________________________

I don't pretend to claim any laurels, so I'll ask you not to touch my handiwork when rating it.

I wrote it in two evenings, it is a beta, what do you want from it? A miracle?

 

Yes, I would like to remind the participants of the conversation that I would still like to hear constructive criticism and suggestions (of any kind) for the product, rather than find out what's cooler.

I am not interested in the latter.


I wrote mine, as I was satisfied with neither the first nor the second version.

 
StatBars >> :

Is the SX(TheXpert) program a combinatorial explosion via MT4?

Or is it still optimising the conditions?

The optimization is genetic, that's why it is written in MQL. The enumeration is accelerated by screening out conditions which are too cumbersome.

 
TheXpert >> :

I wrote it in two nights, it's a beta, what do you want from it? >> A miracle?

No we want the author to share his results. So did you check it yourself after you wrote it? Or do you expect others to do it?


I, for example, wanted to test it, but after a few hours I saw nothing at all, i.e. a complete zero. I had to stop testing. And so the questions are.

 
Reshetov >> :

No we want the author to share his results. So, did you check it yourself after you wrote it? Or are you hoping others will do it?

Well, it's actually working, but actually, yes, I hope so.

I, for example, wanted to test it, but after a few hours I didn't see anything at all, i.e., complete nil. I had to stop the test. Therefore, questions like this.

Turn on the genetics. And if you don't want to turn it on -- bear with me, why, I described it above.

_________________________

I'm a little sick, so I won't be able to respond promptly for a while. I will try to make a version by Monday, in which I will change the order of optimisation, so that there is no initial downtime.

 
TheXpert писал(а) >>

I would ask for another thread to be created to clarify the generator ratings.

I suggest that those wishing to discuss generators and their strategies "pass" Here.


Apologies to the SX author for not using the thread as intended.

And good luck!