EURUSD test with Dukascopy 99% data

Post your ForexRealProfitEA statements here

EURUSD test with Dukascopy 99% data

Postby ronjo » Sun Feb 20, 2011 2:02 pm

I have done the same test with EURUSD.

5% risk with 100 pips S/L.
http://www.palbury.com/EURUSD_5_SL100/

5% risk with 50 pips S/L.
http://www.palbury.com/EURUSD_5_SL50/

Lower profit than EURGBP due to more losses but still very good.
/R
ronjo
Rookie
 
Posts: 3
Joined: Fri Feb 18, 2011 6:07 pm

Re: EURUSD test with Dukascopy 99% data

Postby Support_AdrianS » Tue Feb 22, 2011 10:11 am

Thank you for your research.
There are a few things to say to these tests:
1. The recommended trading time is 21:00 – 21:59 GMT for the summer time (USA DST time) and 21:00 – 22:59 GMT for the winter time (USA DST time).
2. We did our optimizations for Stop Loss and Take Profit without Money Management, because it is more accurate.
3. In the backtests that you made, the profit is higher for the Stop Loss set on 50 because the risk is doubled.
The robot calculates the size of the lots depending on Stop Loss and Risk Percent.
If risk is the same (5%) for both Stop Loss of 50 and for Stop Loss of 100, in the Stop Loss of 50 cases the lots are two times higher than if the stop loss was 100.
This is because you are using the money management.
The profit is higher because the risk is doubled.
4. We do all the tests and optimizations directly on Dukascopy with our java version of the program.

We are glad for your interest and we hope that you will continue this research. We will gladly respond to all your questions.

Kind regards,
Adrian
ForexRealProfitEA Team.
Support_AdrianS
Moderator
 
Posts: 160
Joined: Mon Nov 22, 2010 3:26 pm

Re: EURUSD test with Dukascopy 99% data

Postby magicspider » Fri Mar 25, 2011 7:06 pm

What is "USA GMT TIME" - has the Greenwich moved?
magicspider
Rookie
 
Posts: 3
Joined: Fri Feb 25, 2011 12:22 am

Re: EURUSD test with Dukascopy 99% data

Postby support_CataH » Fri Apr 01, 2011 9:35 pm

Hi magicspider,

It was an error in that post. We meant the daylight saving time (DST) and corrected the post above. Thanks for notifying us.

Regards,
Catalin
ForexRealProfitEA Team
support_CataH
Beginner
 
Posts: 42
Joined: Mon Nov 22, 2010 11:17 am

Re: EURUSD test with Dukascopy 99% data

Postby dambrus » Sun Apr 03, 2011 6:38 pm

support_AdrianS wrote:1. The recommended trading time is 21:00 – 21:59 GMT for the summer time (USA DST time) and 21:00 – 22:59 GMT for the winter time (USA DST time).


Could you comment on results on Birt's EA Review page, indicating better and smoother results for your EA (for all pairs) if running time is 22:00 to 22:59 GMT (all year long)? Are your tick data backtest results different?

Thanks.
dambrus
Rookie
 
Posts: 1
Joined: Thu Mar 17, 2011 12:07 am

Re: EURUSD test with Dukascopy 99% data

Postby support_CataH » Mon Apr 04, 2011 4:30 pm

Hi dambrus,

We don't want to comment Birt's review. He makes great reviews based on deep documentation. But our reccomendations are based not only on backtests, but also on forward tests which had better results from 21:00 to 21:59 GMT. Anyway we only make reccomendations, but you can use any settings you may like. What we reccomend is not mandatory, you are free to use your own settings if you have better results in backtests or if you just want to follow Birt's backtests. It's all up to you. The robot will still be profitable.

Regards,
Catalin
ForexRealProfitEA Team.
support_CataH
Beginner
 
Posts: 42
Joined: Mon Nov 22, 2010 11:17 am

Re: EURUSD test with Dukascopy 99% data

Postby Support_AdrianS » Mon Apr 04, 2011 4:48 pm

We do all the back tests directly on Dukascopy with our java version of the robot.
The accuracy of our back tests is much better than the MetaTrader platform and the tests are very close to reality.

We know about Birt's EA Review back tests and we consider that the tests are very good. He uses the Dukascopy history data as well.
We appreciate his opinion and we believe that Birt is a very good professional in the field.
On our tests, however, there were some different results from those.
On our tests, the trading time from 21 to 22 GMT shows better results than 22 to 23 GMT.
Both trading hours are profitable, but in our opinion trading from 21 to 22 GMT shows 30-50% better results than trading from 22 to 23 GMT (depending on the currency used).

We make only recommendations; nothing prevents you from using both hours of trading. Only a forward test on a live account on long term can confirm or refute the claims.

Thank you very much for your interest and for your very good question.

Kind regards,
Adrian
ForexRealProfitEA Team.
Support_AdrianS
Moderator
 
Posts: 160
Joined: Mon Nov 22, 2010 3:26 pm

Re: EURUSD test with Dukascopy 99% data

Postby muzzamcc » Thu May 05, 2011 2:08 am

For what it's worth my tick data backtesting using dukascopy data and reasonable/high fixed spreads is showing that 21-22GMT year round is better than 22-23GMT year round. Lower drawdown % and higher profit figures. I can't work out how Birt got the results he did, unless it was just that the robot he tested was a much earlier version than what i have just tested with. (Demo version 5.23)

Regards,

Murray.
muzzamcc
Rookie
 
Posts: 4
Joined: Sun Mar 06, 2011 8:31 am

Re: EURUSD test with Dukascopy 99% data

Postby Support_AdrianS » Fri May 06, 2011 9:36 am

Dear Murray,

We believe that Birt's results are much different for two reasons:

1. Birt used a much older version of the robot when he made these tests and since then many improvements have been made.

2. We just discovered a problem yesterday, but only when testing on Dukascopy historical data with the MT4 platform.

The Spread in MT4 version of our expert advisor is calculated by the formula:

Spread=MarketInfo(Symbol(),MODE_SPREAD);

In forward test, when the program runs in real conditions, the spread is calculated correctly.

In back tests on MT4 platform the spread is also calculated correctly, because historical data do not include historical data for spreads and the test is done with the spread of the testing time.

On our java version of the robot for Dukascopy, the spread is calculated by the formula:

Spread = Ask – Bid;

We make all the back tests and optimizations directly on Dukascopy with our java version of the robot and because the spread is calculated as the difference between Ask and Bid the tests are performed correctly and the problem that we mentioned does not appear.

We will add to the MT4 version of our expert advisor an additional condition in order to properly test the robot on Dukascopy historical data on MT4 platform also.

This addition will be available at the next update, hopefully next week.

Kind regards,
Adrian
ForexRealProfitEA Team.
Support_AdrianS
Moderator
 
Posts: 160
Joined: Mon Nov 22, 2010 3:26 pm

Re: EURUSD test with Dukascopy 99% data

Postby muzzamcc » Sat May 07, 2011 8:17 am

Hi Adrian,

Thats a pity, looks like my backtests will all have to be redone, just when I had finished them too! Just a suggestion, why don't you put proper tick backtests on the webpage rather than the 90% backtests you have there, or maybe in addition to them. Might be worth letting Birt know that his tests may not be accurate too. He trades your robot from 22-23 GMT based on his tests and that might not be the best time. You may be missing out on sales that he could be generating for you if his forward test was doing better.

Keep up the good work,

Murray.
muzzamcc
Rookie
 
Posts: 4
Joined: Sun Mar 06, 2011 8:31 am

Next

Return to Statements

Who is online

Users browsing this forum: No registered users and 2 guests

cron