WF matrix and optimal complexity

In ‘Trading Systems’ by Tomasini and Jaekle, they mention the concept of a system’s optimal complexity. For example, if there are 5 optimizable parameters, you should only optimize 2-3. Beyond that you are likely to suffer OOS deterioration.

In SQ’s WF matrix, you can optimize as many parameters as you wish. In order to determine the most important parameters to optimize, I suppose you could run the WFM several times with different parameters being optimized, and select the best results. Does anyone have experience with such a process?

From what I have read (pardo & davey) the WFO requires only 2-3 parameters to be optimised as you noted. In order to find the best parameters to optimise simply run WFO with only one parameter chosen, complete all other settings as normal. Run the WFO (should only take a couple of minutes) and log the results of the net profit / RDD / pf and other fitness function you rank highly. Follow this process for all of the parameter candidates and apply only those 2-3 parameters into your full WFO run with the highest variance.

With regards WFM, I don’t follow common theory as SQ allows for genetic simulation instead of brute force, I apply all parameters to my WFM runs and use 30 or 50% variation depending on strategy.

Sent from my iPhone using Tapatalk