Menu
Sister Sites

Options

Bookmark Thread

First Page First Unread Last Page Last Post

Print Thread

Similar Threads

multicurrency datas with OnChartEvent() and EventChartCustom() 24 replies

MT4 multicurrency EA data problem 5 replies

MultiCurrency Candles 3 replies

Chamunda's MultiCurrency Strategy 237 replies

Multicurrency ea for demo testing 1 reply

Backtesting multicurrency strategies?

  • Post #1
  • Quote
  • First Post: Nov 12, 2007 7:10am Nov 12, 2007 7:10am
Hi All,

As it is known, MetaTrader platform doesn't allow to backtest multicurrency strategies. Do you have any proven methods to calculate such values as maximal drawdown or profit factor for multicurrency strategies? Is there anything better than mundane summation of numbers from each report?

Best regards,
Vitovd
do you mean like pseudo hedge strategies?
Not necessary pseudo hedge strategies. I mean more generally, strategies that trader wants to run on several pairs rather than one to make his equity curve smoother. And the basic question is: is the maximal drawdown much lower from all pairs than from each pair separately.

Best regards,
Vitovd

http://www.forexfactory.com/images/misc/progress.gif

You could take a look at Rightedge. It supports portfolio level backtesting, multi-currency support(including historic exchange rates to calculate correct p/l) and also accounts for historic rolls (using the oanda interest rate history). Its pretty good, coding is done in .NET (vb.net or C#).
RightEdge looks quite powerful but it doesn't support MQL language and doesn't have interface to MetaTrader platform for trading. So any strategy or indicator should be ported to C# only for testing.

I'm looking for some tool that can help in analyzing the performance of multicurrency strategies written for MT platform. If there isn't any I'll probably have to create a simple tool that aggregates reports from multiple backtests. It’s the only way, I think. It won’t support optimization but it will be better than nothing.

Best regards,
Vitovd
Quoting Vitovd
If there isn't any I'll probably have to create a simple tool that aggregates reports from multiple backtests. It’s the only way, I think. It won’t support optimization but it will be better than nothing.
That's probably the only way unfortunately. There are many great portfolio backtesting platforms, but none that I'm familiar with that will inter-operate with MT4.

The ET forums have hundreds of threads on these various platforms, so I'd search there as well if you're looking for additional info. Here's a recent thread of platform reviews.
Thanks for the link to ET forum. However, I’ve decided to create simple tool that aggregates MT reports. If there was an interest in this program it could be available to a broader group of people.

Best regards,
Vitovd
There was an article at mql4.com about this but can't find the url atm
The program to aggregate MT reports is almost ready. Let me know if someone is interested in it. It will be available as freeware software. And if the level of interest is quite high maybe I’ll prepare small user guide (although the interface is quite simple).

Best regards,
Vitovd
Quoting Vitovd
The program to aggregate MT reports is almost ready. Let me know if someone is interested in it. It will be available as freeware software. And if the level of interest is quite high maybe I’ll prepare small user guide (although the interface is quite simple).

Best regards,
Vitovd
Heavens ! A dream come true for me........

I'm into heavy metal backtesting of 9 pairs using a specific ea (tested best time frames, optimal SL, TP, TS, DMI and MOM) for each of 9 pairs over a period of 9.75 months and am about to finish a round of new settings.

The last time I "compiled" the data to get an overall picture, what I did is I copy-pasted the report for each pair for 9.75 months into a separate Excel spreadsheet, then in a "main" spreadsheet (yes, it's crazy but true) I went on to "intertwine" all trades (there were a gazilion of course) in a chronological fashion to simulate what an integrated report would look like. I then made some monthly calcs, then general calcs and finally a graph of the whole thing (I also did this separately for each pair to get a glimpse of how individually they performed, when their weakest periods were, etc.) The "general" report helped me figure out if at any point the account went under, by how much, for how long, etc.

This took a lot of time, as you can imagine, and right this saturday, exactly now, I was about to generate the first report of the group to start this whole process again.

And then I read your thread AMAZING!!
So, am I interested ?? You betcha! X 1000000000000

I'm also posting my testing results in this thread, if ever you're interested:

http://www.forexfactory.com/showthread.php?t=45110

I'm going to pm you right this minute
I am very interested also.

Look forward to seeing it in action.

Thanks
OK. I've just uploaded the program to my website and I've added some description. You can download it from here: http://www.mqlsoft.com/download/programs. Please note that this is the first version and it is very simple but I hope it will help you. The program requires Java 6 (Java Runtime Environment) to run.

Best regards,
Vitovd
Quoting Vitovd
OK. I've just uploaded the program to my website and I've added some description. You can download it from here: http://www.mqlsoft.com/download/programs. Please note that this is the first version and it is very simple but I hope it will help you. The program requires Java 6 (Java Runtime Environment) to run.

Best regards,
Vitovd
Hi Vitovd,

Thanks, downloaded and tested (last week actually).

First off, great tool, congrats on your programming.

One thing though, I actually thought it could merge the “individual trades” like I do in Excel, that is, intertwine the trades from various pairs in a chronological order and provide a “statement” of the trades as if taken in time one after the other. This way, I get to calculate monthly performance “as a group”, etc. But that’s ok, I’ll continue doing it by hand (good to practice patience hehe)

However, your program does provide the end result, ie, the graph which is merged. I posted some results in the Supernova journal, if you’re interested.

There’s one thing that isn’t clear to me however. I’ll explain. Let’s say my account has an initial deposit of $5,000. Now, I’m trading several pairs (9 in my example) all in mini lots and all are from this same account, not separate accounts. So the merged graph’s initial deposit should always be $5,000. But with your system, as I merge the first 2 sets, the new deposit is now $10000 (and so on) along with the balance. Now, I’m worried that this doesn’t give accurate Profit Factors, Drawdown, etc. (but of course I could be wrong). Can you clarify this point?

Thanks again, and a MERRY CHRISTMAS !!!!!!!!
Quoting Qpip
One thing though, I actually thought it could merge the “individual trades” like I do in Excel, that is, intertwine the trades from various pairs in a chronological order and provide a “statement” of the trades as if taken in time one after the other.
Hi,

In fact ReportManager does what you've written, but it stores it in an internal memory. So the only problem is to present the data.

Quote
There’s one thing that isn’t clear to me however. I’ll explain. Let’s say my account has an initial deposit of $5,000. Now, I’m trading several pairs (9 in my example) all in mini lots and all are from this same account, not separate accounts. So the merged graph’s initial deposit should always be $5,000. But with your system, as I merge the first 2 sets, the new deposit is now $10000 (and so on) along with the balance. Now, I’m worried that this doesn’t give accurate Profit Factors, Drawdown, etc. (but of course I could be wrong). Can you clarify this point?
The program sums initial deposits of all reports. At first I thought that it is the best solution for reports with the different size of an initial deposit. But you are right. If someone is sure that his strategy will have enough money to trade the program should have an option to set the value of initial deposit for merged report.

These two points will be taken into consideration during works on the new version.

Best regards,
Vitovd
The new version for the New Year is available. What's new:

  1. Possibility to save reports to .htm and .csv files.
  2. A new Statement panel with the list of all orders.
  3. Ability to change some settings in the Preferences.
  4. Reading reports in languages other than English.
  5. Close All Reports option.
  6. Some bug fixes.

I've created the alias to download page: http://reportmanager.mqlsoft.com/.

Best regards,
Vitovd

This is an incredibly useful software, thank You for releasing it.
congratulations!
Rather late, but a big thanks from me as well

Later all..
Hi,

Can you add some money management routines? Like trade lots = Equity/2000?

Thanks
  • Post #19
  • Quote
  • Last Post: May 27, 2008 6:00am May 27, 2008 6:00am
Quoting Scientist
Hi,

Can you add some money management routines? Like trade lots = Equity/2000?

Thanks
Hi,

Thank you for the interest. But I don't think this is the most important feature to implement. Because if you already have a profitable EA it will only answer the question how fast you will become a millionaire. Moreover the program would have to change original volumes of trades. For each currency pair the same amount of lots requires a different margin so the best option is to leave volumes of trades unmodified.

ReportManager will rather go in a different direction - to find as much weak points of EA as possible. So in a new version I will add more performance factors because only Profit Factor is insufficient. Then program will be useful not only for trades who test multicurrency strategies. I don't want to make too many promises now because I have to at least find some time to program it. But I have a few ideas which I think are very interesting.
0 traders viewing now