help me feed my data-hungry EA

 

Ok, so I've developed some EAs which perform statistical categorizations of market behaviours and then produces monte carlo predictions that drive trading decisions based on the calculated statistical data. It should work on any currency pair, since the stats are calculated on the fly for whatever chart it's loaded on, so whatever behavioral differences between pairs are already baked in.

It sounds great in theory, but there are some serious real-life brick walls i'm runing into. Because it uses chart data to calculate the stats, and because it needs sufficient data for each different "market state" (there could be hundreds or thousands of states), I need an ungodly amount of data loaded in my charts. I'm talking millions of bars in order to have meaningful statistics.

I already maxed out the "max bars in history/chart" settings. I held down the "home" key and got a bit more data ~(60k bars), but this is nowhere near enough. I think i read somewhere here that you could change one of the datapoints in the history center to 1970 and it'll try to fill the hole with downloaded, but that didn't work. I tried to find that post again but couldn't. Does anyone else have any ideas?

Thanks,

 

You are probably referring to my post here: https://www.mql5.com/en/forum/124619

You can also try the links in my post here: https://www.mql5.com/en/forum/118275

Any chance you'll be sharing some of your source codes with the forum?

 
you can also try to directly access the .hst files in the history folder. The file format is quite easy, it is documented in the period converter script. But you need to download and/or import the data from somewhere first.
 
produces monte carlo predictions that drive trading decisions based on the calculated statistical data.

Yeah I was just curious to see how one goes about setting this up because I had something very similar in mind. I have a MonteCarlo routine coded but haven't done anything to make it do future projections and so on, that's me next project. That's what piqued my interest about your post, but I can understand if you'd rather not let this become public.
Reason: