Hello,
I have a Signal. It curve fits pretty significantly so after an optimization. of the 100's or sometimes 1000's of possible settings the optimization generates only about a 1/3 of them test well in OOS data.
This is troubling for sure.
BUT
Does it sound reasonable to run an optimization then plug in a significant portion of the Variations into copies of the signal inside a larger "Master Signal" then create an array of change in Net Equity of each of the Variation and have the "Mast Signal" select the top candidates and trade them all?
Is this possible?
Is it going to hammer my computer?
I have a new Dell with
16 gigs of Ram,
I7 (8th gen) processor,
238 Gig SSD
Have any of you done this?
In Signal Optimization
Re: In Signal Optimization
It is doable in backtesting buy setting the trade size to 1, but would be inefficient (and with any exchange-traded product with some decent min.lot size - impossible) to implement in live trading with PT.
Unless of course you implement all systems as functions and calculate/maintain all needed statistics yourself.
Please read and VOTE:
https://www.multicharts.com/pm/public/m ... es/MC-2367
Unless of course you implement all systems as functions and calculate/maintain all needed statistics yourself.
Please read and VOTE:
https://www.multicharts.com/pm/public/m ... es/MC-2367
Re: In Signal Optimization
Yes.
It is worth reviewing it from time to time and casting your ballot/filing issues - if you want to change things.
And it will improve your understanding of how MC/PT work (or do not work) immensely.
It is worth reviewing it from time to time and casting your ballot/filing issues - if you want to change things.
And it will improve your understanding of how MC/PT work (or do not work) immensely.