Best settings for Genetic Evolution

Hello guys,

What are your Best settings for Genetic Evolution ?

Example :

• Population size : 100
• Number of generations : 100
• Initial population factor : 2
• Mutation probability : 25%
• Restart fitness if stagnates for : 5 generations
• Initial population screening conditions : Net profit <0 ; Nb trades (OOS) <20
• Size of the databak : 2000
• then Ranking  options for screening of strategies :
• Profit Factor IS and OOS < 1.3
• Return/DD IS and OOS < 2
• DD% IS and OOS >60%

Hello,

there is nothing wrong with the settings. Maybe two notes: high number of generations can produce too many similar strategies (also depends on tree depth and crossover setting). Increasing the population size generally adds more variety into the generation process

In my opinion, there is no best setting. It depends on the type of strategy problem with which you are working. Your example is a good starting point but there are times when 100% mutation, 100% cross-over with a stagnation of 50 are appropriate. The strategy development research question ultimately drives the GE settings.

Hi notch,
I’m still reasonably new to SQ so still getting my head around concepts. I’m interested by what you said, would you mind elaborating with as much detail as you can about any processes for determining what type of problems suit different evoultion settings and the components of those problems that define them? i.e Is it defined by desired outcome, or indicators used, or markets etc.

thanks, Al

Enclosed my current settings. Opinions are welcome

(The Filter generated inital population is variabel, depend on the timeframe)

May i know why occur this error while building strategies?

You made settings in the “genetic options” field, right? They are apparently already so strict that SQ cannot create an initial population.