History Center problems and how to use properly?

 

After playing around with EA's for a bit, and using the strategy tester mainly to see if the code was actually working I'd now like to progress onto the real why of the history center, and how to use it properly and I can't seem to find the answers or something comprehensive on it.

OK I get mismatched chart errors on an EA test, and I've seen posts on getting rid of them using the history center, and the history center being "bogus" data. This raises questions for me.

If I run an EA in visual mode inside the strategy tester I can see it entering when it is supposed to be. Running all the data and sitting patiently shows nothing going wrong. So how is the error message of mismatched chart errors effecting my results? What exactly is a mismatched chart error?

So to get a better set of results, I download the data inside the history center. So lets take for example the chart below on EURUSD. That gap appears due to the download center data? So to get rid of it I have to delete the history records and get it to download a fresh batch ( from GFT in this case) so I can go back to trading live again with good data? Would it be wise then to use a development MT4, from a broker which I won't hold an account with and do my EA development inside that MT4, and transfer it to another MT4 for live testing. And download history center data for testing all my EA's?

So how do you use this history center effectively? What about the gap in the data I have below? Do you guys download data from the history center, and then disconnect from the network whilst you are running the strategy tester so you don't get this sort of stuff happening? What about spread when you are not connected?

.

Cheers

Colin

 

IMO, (Problem Solve) is better than (Why)
1) Disconnect from your broker by using a bogus proxy.
2) Delete all your old data.
3) Download data from reliable source.
4) Use period converter script to create all periods.

Now my attempt at your Why. As I understand it.
I'll start with pointer 2) first.
2) Old broker data and New_Downloaded data will conflict causing mis-match errors.
1) If you don't use a bogus proxy then every-time you open mt4 the broker will replace the Deleted data.
3) History center download button does-not give reliable data as-far-as-most here are concerned due to the gaps.
4) Because if you don't you'll only have 1-minute timeframe data.

For your other questions:

So how is the error message of mismatched chart errors effecting my results? It does not.

What exactly is a mismatched chart error? The 1-Minute data from the reliable source you downloaded shows EUR/USD opening at 1.5000 at 2_am. The H1 bar opening at 2_am from your broker however shows price opening at 1.5001. Thats Mismatch error.

That gap appears due to the download center data? Yes, and downloading a fresh batch wouldn't help. You'll get the same data with the same gap. Best case scenario, you want to download from you broker server by using the Scroll Back <- arrow or page-up/page down buttons. But the broker's m1 data is limited in time. Therefore we find ourselves looking for other sources who will give us longer years of data (without the gaps of course).

So I can go back to trading live again with good data? Most people here will recommend having 2 mt4-folders, biscally you copy the original folder. You keep one online and one-offline (by using the bogus proxy method). You keep your development environment in the Off-line along with your Downloaded Data there.

a broker which I won't hold an account with...? No, and again don't use History center's data (unless you have no other choice). The back-testing environment should look as much like the broker you intend to use it on. Because things like Stop_Level etc. Could make the system look more or less attractive. You don't want to get to live testing to realize its not compatible.

So how do you use this history center effectively? For importing and exporting data. Not for downloading data. Its also decent for managing data.

What about the gap in the data I have below? What about it... Its there and will remain there. The data it self is not from MetaQuotes . But rather it's from which ever broker MetaQuotes partners with as the promotional data provider. Reason why this have not been fixed beats me. But traditionally quality data in Trading has been something people paid for so maybe blah.

What about spread when you are not connected? Couple of things about the back-tester. When you hit that Start button. The back-tester checks the 1-Minute data against whatever time frame you're testing for like H1. If M1 and H1 dont match then you get an Warning-mismatch error and then it proceeds with the test. When you are Not connected, then whatever spread it had at the time of disconnection is what it'll keep forever or til connected again.

Also it looks at the Current market properties like spreads at the moment you hit the Start-testing button. If it's the weekend any your testing platform is connected to your broker and your broker is open on the weekends and changes the spreads to something like 20_pips during that time, then it's going to make results look different than if that same test was ran on monday morning.

Anyway, I'm not going to do all the work for you here. You'll want to search the forum for stuff like. Bogus proxy, Period_Converter and Data Sources. I've posted my favorite free provider allot of times.

 

Much appreciated for the explanations and especially the search handles to find out more. Looked at your posts and didn't find your best free provider, but the search is on now to get this strategy tester running a bit better. Shame there isn't one full article on it somewhere, everyone has a little bit of good info :(

Cheers

 

Looked at your posts and didn't find your best free provider...

This is the one I was referring to. I also like these.

 

Here is an article about mt4 backtesting with quality data

Reason: