Download MetaTrader 5

Backtest Market Data and the real chances

To add comments, please log in or register
chinaski
921
chinaski  

Hello guys,

just was about to make a backtest for German DAX and was disappointed about the poor data provided.

My broker (ActiveTrades) told me, sorry, but this is best data quality we have. If you need more, try Equity Clock.

 

Did not find any data on equity clock.

 

Also, what should be the sense, to use extern data instead of those coming from my broker ? Because based on those data i get fills.

A backtest would not have a good evidence, if using extern prices.

 

 

I mean, backtest without realistic data makes no sense.

And automated trading without backtest makes no sense as well.

So there is a big dependency of a successfull auto trading  from accurate data.

 

How do you deal with all those inponderabilities ? Do you pull external data ?  

If yes, are recommendations possible ? 

Does your broker serves with better data ?

 

Thank you for comments 

Marco vd Heijden
Moderator
5336
Marco vd Heijden  
chinaski:

Hello guys,

just was about to make a backtest for German DAX and was disappointed about the poor data provided.

My broker (ActiveTrades) told me, sorry, but this is best data quality we have. If you need more, try Equity Clock.

 

Did not find any data on equity clock.

 

Also, what should be the sense, to use extern data instead of those coming from my broker ? Because based on those data i get fills.

A backtest would not have a good evidence, if using extern prices.

 

 

I mean, backtest without realistic data makes no sense.

And automated trading without backtest makes no sense as well.

So there is a big dependency of a successfull auto trading  from accurate data.

 

How do you deal with all those inponderabilities ? Do you pull external data ?  

If yes, are recommendations possible ? 

Does your broker serves with better data ?

 

Thank you for comments 

I agree that a backtest without realistic data makes no sense.

However i disagree that automated trading without backtest makes no sense.

I rarely use the backtester and if i do i use MQ data.

Fernando Carreiro
2321
Fernando Carreiro  
chinaski: just was about to make a backtest for German DAX and was disappointed about the poor data provided.

If using MetaTrader 4, you can use 3rd Party tools and data (such as TickStory and GER30 tick data from DukasCopy) in order to produce HST and FXT files for back-testing. However, on MetaTrader 5, things are more limiting!

In my case, before writing MQL code, I first write a "C/C++" version of the strategy (Microsoft Visual Studio 2015) and test it directly against the native Dukascopy Tick data before trans-coding it into MQL. I have setup my C/C++ code library to mimic the MQL environment as close as possible so that trans-coding it, is relatively simple once I am happy with it.

chinaski
921
chinaski  
Fernando Carreiro:

If using MetaTrader 4, you can use 3rd Party tools and data (such as TickStory and GER30 tick data from DukasCopy) in order to produce HST and FXT files for back-testing. However, on MetaTrader 5, things are more limiting!

In my case, before writing MQL code, I first write a "C/C++" version of the strategy (Microsoft Visual Studio 2015) and test it directly against the native Dukascopy Tick data before trans-coding it into MQL. I have setup my C/C++ code library to mimic the MQL environment as close as possible so that trans-coding it, is relatively simple once I am happy with it.

Thank you fernando for this suggestion,

but so far i prefer MetaTrader 5 because of it's position handling and transaction event, for various reasn.

ActiveTrades offers MT5 to use DAX with very low funds.

 

But what you say sounds very interesting: Does it mean you get free tick data from DukasCopy ? 

 

 

VS2015: I can't see a benefit here: If you stick with dukascopy, what should be more directly than backtester backtesting on dukascopy data ?

Thank you for clarifying. 

Alain Verleyen
Moderator
32814
Alain Verleyen  
chinaski:

Hello guys,

just was about to make a backtest for German DAX and was disappointed about the poor data provided.

What do you mean by poor data provided ?


How do you deal with all those inponderabilities ? Do you pull external data ?  

If yes, are recommendations possible ? 

Does your broker serves with better data ?

 

Thank you for comments 

You can't use external data with MT5 yet, but it's promised by Metaquotes. No idea about the deadline though.
Fernando Carreiro
2321
Fernando Carreiro  
chinaski: Thank you fernando for this suggestion,

but so far i prefer MetaTrader 5 because of it's position handling and transaction event, for various reasn.

ActiveTrades offers MT5 to use DAX with very low funds.

But what you say sounds very interesting: Does it mean you get free tick data from DukasCopy ? 

VS2015: I can't see a benefit here: If you stick with dukascopy, what should be more directly than backtester backtesting on dukascopy data ?

Yes, Dukascopy data is free!

The benefit of using  VS2015 (or any other C/C++ compiler) is because importing 3rd party tick data into MT5 is a extremely difficult if not impossible. So, the main benefit, is that I don't have to convert the tick data and can test against it directly and also because my back-testing (and/or optimisation) is extremely fast (much faster than MetaTrader 5 tester). I can also simulate things like slippage, variable spread, network delays, outages, etc. in order to make sure my code is robust enough.

chinaski
921
chinaski  
Fernando Carreiro:

Yes, Dukascopy data is free!

The benefit of using  VS2015 (or any other C/C++ compiler) is because importing 3rd party tick data into MT5 is a extremely difficult if not impossible. So, the main benefit, is that I don't have to convert the tick data and can test against it directly and also because my back-testing (and/or optimisation) is extremely fast (much faster than MetaTrader 5 tester). I can also simulate things like slippage, variable spread, network delays, outages, etc. in order to make sure my code is robust enough.

@Alain:

poor means (75 % missing). The backtest result indicates this:

 

@Fernando: 

Okay, when backtesting with MT5 and you can't import external data, it makes sense. But with MT4, there is no need to take the detour via vs2015 or don't agree ?

Thank you 

Fernando Carreiro
2321
Fernando Carreiro  
chinaski:

@Fernando: 

Okay, when backtesting with MT5 and you can't import external data, it makes sense. But with MT4, there is no need to take the detour via vs2015 or don't agree ?

Yes, that is why I gave you instructions on how to do it in MetaTrader 4.

However, I write code for both MetaTrader 4 and MetaTrader 5 and my environment is setup so that the code I write works on both (I have a my own API library). This allows me to write code only once (VS2015 in C/C++) that compiles and works in all 3 of my environments, MQL4, MQL5 and my own C/C++ setup.

So, with this in mind, it is no longer a "detour" and I can test and run in all 3 environments, and if one of those, namely my own, can offer me better testing tools, that's the one I will use most.

Obviously in your case, you will do what best suits you. I just described my setup, because it could spark an idea for something similar in your setup, if you wanted to pursue such a solution of your own.

chinaski
921
chinaski  
Fernando Carreiro:

Yes, that is why I gave you instructions on how to do it in MetaTrader 4.

However, I write code for both MetaTrader 4 and MetaTrader 5 and my environment is setup so that the code I write works on both (I have a my own API library). This allows me to write code only once (VS2015 in C/C++) that compiles and works in all 3 of my environments, MQL4, MQL5 and my own C/C++ setup.

So, with this in mind, it is no longer a "detour" and I can test and run in all 3 environments, and if one of those, namely my own, can offer me better testing tools, that's the one I will use most.

Obviously in your case, you will do what best suits you. I just described my setup, because it could spark an idea for something similar in your setup, if you wanted to pursue such a solution of your own.

I appreciate this. Understand your point now and it makes sense.
To add comments, please log in or register