Built objectives in genetic module

I am interested in all comments, even the most critical…..

 

Lets consider the genetic method, .  

It starts by “seeding”, i.e. creating random set of initial  strategies.   In the setup I see the option to set the objectives: Settings>Genetic options > > Initial population conditions.  I have an ambitious plan to come up with a strategy that will approximate trends on a daily data.  Thus I am less concerned with, for example,  low win% parameter, and more concerned with higher win to loss ratio.  I also aim for decent expectancy (nothing fancy here, its just a seed).  At the same time I don’t want in that initial population any negative returns in the out of sample part of the data.  I consider such restrains reasonable.  

Upon start the program attempts to build the initial population.  However, if it takes more then predetermined number of iterations to do it, the program pauses, and displays error message : “can not generate valid initial candidates. ”  

So far I determined that the number of restrictions and not their value plays a pivotal role in the “seeding” of initial population.  

Why are the restrains important? The bad “seed” will have a strong tendency to converge to equally bad strategies.  This is my experience so far. after using the program for several months.

By trial and error, playing with both the number of objectives and their values, I was able  to force the program to create a few random strategies that looked decent .  I tried to “harvest” them, i.e., to save them and collect a decent number of them to start the evolution, but I don’t think it is possible.  

 

The initial seed for genetic module seems to relay on random engine.  Why can’t it go longer, as long as it takes to populate the “seed”? 

Alternatively, why cant I “harvest” several strategies from different attempts to seed and deploy them for “evolution” ? 

 

 

You can.
Do random of lets say 1000 or a million (store them in the bank) put rules that all these strategies must have is and oos netprofit > 0
With win rate bla bla.
Stop random generation when satisfied .

Now go with genetic (before that save the random strategies just as a safe side).
Go to genetic settings and choose genetic from bank .
Thats it

Got it, thanks

Ok, but genetic algo should try to converge to built objectives.  It doesn’t.  When the number of built objectives is >2 the the program fails to create the seed in genetic mode.  From what I understand, the seed should be random and the algo supposed to converge to objectives.  If I understand the objectives, i.e., I know what kind of properties I expect from the built, the program should be able to help me.  Now, because of the fact that the initial generation is random, the “evolution” may fail, and that’s OK, I may try it again.  Unfortunately it doesn’t happen that way.  

The alternative ( random generation with constraints) is doomed to failure from mathematical point of view. 

random wont fail you , many here use only random . theories are just theories .

if you want to generate smth that u feel work the best you can do the following . 

create strategy in sq3 lets say 

 

buy if rsi > 60

sell if < 40

 

put sl and tp based on atr or fixed .

 

now  go to improve strategy section and put (add to first rule and second) you will have genetic with the spirit of strategy that u looking for