Genetics spam

Sq generates thousands of the same strategies

How to diversify them without using random option?

increase mutation probability . but depends on more factors. it is always better to post here your settings and not let others to guess

good idea , will try

Simply you can select higher amount of building blocks being actively used. When only few are selected, as the time passes your strategies obtained will tend to converge since you have limited space of possible setups

SQ finds a good params and sticks to it

thus thousands of similar strategies param by param

can you stop SQ to spam and do more diversification

lets say if found 100 strategies that fit params then move on to new ones with new indicators

SQ strategies are lacking diversification… literally 9000 strategies which are the same

I have no problem like this. So you have to improve your setting and it will work

How to improve

Give me tips please

without your building settings no one will tell you anything – post here something

i am using SQX for many months and i have no problem with diversification – it looks like to me, that you dont know what you are doing…

yes, I am learning, still the same problem, tried your ideas, did not work

can you show me your params?

how do you diversify strategies?

its better to start with your settings, it will be better – its not effective to copy someone – you need to find your own way

Make sure to use lots of building blocks as suggested

Also use 1 to 3 or 1 to 4 conditions for entry and exit to add even more diversification

Either use very strict ranking or limit databank size to a low number and generate strategies for a period of time where databank fitness will continue to improve as older strategies get replaced with newer strategies.

I’ve attached some example screenshots from a project I’m using at the moment that is providing some nice diversification

I would also suggest you click on the help button in the genetic options menu and have a good read of that to help you set this up correctly.

The other thing to do is experiment and build your settings up based on trial and error

Ive had success in using random gen and back testing them on previous data. they perform on such data i use them in genetic engine and decimate them to find good un correlated strategies that pass all tests

random brute force is not working

it takes 6 days of searching at 3 million strategies per hour to get 0 strategies with filter pf 1.3

gin wrote:

random brute force is not working it takes 6 days of searching at 3 million strategies per hour to get 0 strategies with filter pf 1.3

really? i dont trust your opinion, because for me it works – you are doing something wrong, because simple filter for profit factor is nothing, this is an easy one