Originally posted by Zestfulmocha
Nikl, Sharks has just referred you to his dodgy mate in the back alley to go buy a kidney - so take this all with a grain of salt as I'm no expert and have only used metaquotes data so far in my testing, and haven't even got an EA live yet lol, so I have no credentials clear.png
If you're really interested you can try QuantDataManager and download quite a high quality Dukascopy data sets (dating back to the dinosaurs) for both tick and M1 (which you subsequently can convert into whatever you want obviously). Here is an article you might find interesting -
https://forums.babypips.com/t/how-i-get-99-9-modelling-quality-on-mt4/64592 and references the platforms I just have.
The download is free, but it's SLOW - there is a paid option with non-capped download speeds but I was in no rush so just downloaded it the slow way. I've just moved house so actually haven't had the chance to complete the downloads on all pairs (moved in 11th, NBN doesn't connect until 27th... what a joke) but I think I have three of the majors completed so I'm ready to run them through MT4 soon - my plan is to run a comparison on a few different models to see what difference the data makes to the results. Stupidly I downloaded all of the M1 data then realised the platform wouldn't convert the files to Mt4, only tick converts to the MT4 requires history format for some reason. There may me ways around that but I haven't looked into it yet.
Rather than try download this, if I notice a different in results, I am happy to find a way to share the data files directly with you.
Also worth noting, for me personally 90 or 99 means nothing, even if there is a difference in results. Let's say 99% showed a 9% drawdown, but 90% showed an 8% - your first reaction is what, 99% is better because it exposed more? Well, what if the live demo results only show a 7-8% draw down in the first year of running it? Who really knows, maybe 90% was better to begin with lol. Not an exact science unfortunately, and frankly even a 100% accurate back test means only so much compared to live results (not telling you anything you don't know sorry) so for me I'm not losing sleep over it, was more of an interest I just wanted to flesh out the answer more than my belief it will have any material impact.
To answer your question more directly, whether you have 90% using the meta quotes data or 99% I am skeptical as to how much of a difference it really makes - perhaps it makes a few % difference, but the difference between the two results (if any) is going to small in comparison to the difference between back test / live demo results anyway so perhaps it's time not well invested at this stage in your EA career? Advice I should probably take myself
Only other thing I would say is if you are going to be doing any more complex stress testing of your model outside of the standard MT4 back test environment, you might want cleaner data as the more you hammer it the more those discrepancies are going to show, but I don't think anyone here is at that level (I'm certainty not).
P.S. - What EA Builder are you using?
P.P.S. - If you're not scalping or HFT, everything I have said above is even more true in my opinion, if you're doing a H4 system for example, the tick data is probably going to add nothing over the bar data sets as any actual deviation is going to be more insignificant to your result the higher time frame you go. M1 to M5 systems? maybe... M15+ I think don't worry.
P.P.P.S. - Sorry all I have not been participating or even reading the threads lately I have been so busy. Will play catch up soon and be back to annoy folks before you know it.
Hey Zestfulmocha
thanks for the link. A good article on how to import tick data to MT4 which I might be able to follow!
Which EA builder do you use?
cheers
