development process

Hello all,

 

I’ve been testing SQ for a few weeks now and have a couple questions. Which one do you find more effective strategy development process?

 

As found in SQ website articles: 

1) Generate (quick generation process) large list of strategies on on high timeframes and loose ranking conditions

2) Retest/filter out on low timeframes and strict ranking conditions

3) Pass via robustness filter.

4) improve and optimize if found something promising

5) repeat from the step 1 if nothing useful

 

As found in Zdenek ebook:

1) Generate (slow generation process) list of strategies based on low(precise) timeframes,

2) Retest/filter out strict ranking conditions

3) Pass via robustness filter.

4) improve and optimize if found something promising

5) repeat from the step 1 if nothing useful

 

I’m suspecting that the only difference will be interaction time required with the application.

 

Other proven process, that you may want share or discuss?

 

Another question does it (or when) make sense to put strict rules upon Initial population conditions? Or just min. number of trades is fine in 99% of cases?  

 

I’ve tried enforcing strict conditions for initial population though it takes ages to get even 200 pop. size.

 

Regards

 

 

Thats what i found as well. Setting up tight rules for initial population makes the overall process a lot slower. I am going for a very easy initial filter (actually its minimum of trades only) and then let the genetic population taking it from that point to improve the strategies. To counteract this, i do have restrictive filters for the databank, because i only want to list strategies that are worth to look at. Otherwise it just floods your databank with useless data.