1 (edited by yonkuro 2016-05-29 06:22:50)

Topic: Different Backtest Result using Percent Change Indicator

Hi Popov,
Could you please check the attached strategy. This strategy only uses one indicators, Percent Change. It shows different result in FSB and MT4 Backtester, the MT4 data is from Dukascopy
http://s33.postimg.org/qoaj9yba3/SS_29052016.jpg
http://s33.postimg.org/8goq61imz/SS_29052016_2.jpg

Regards

Post's attachments

tes 28052016.xml 11.58 kb, 2 downloads since 2016-05-29 

You don't have the permssions to download the attachments of this post.
I’m working on improving my English and trading skill

Re: Different Backtest Result using Percent Change Indicator

There is no problem with the indicator.

I also received a result like your in MT and found out that it doesn't closes the orders properly on the test.
Try reducing the Bar Close Advance interval of the expert from 15 to 5 seconds.

http://s33.postimg.org/j2wmyazrf/screenshot_1677.jpg

With that change and fixing the spread, I received much closer result.

Re: Different Backtest Result using Percent Change Indicator

Thank you for your reply Popov, I'll try that solution.
BTW, may I know why don't make 5 as default value for Bar Close Advance interval paramater as it gives closest result to MT4?

I’m working on improving my English and trading skill

4 (edited by yonkuro 2016-05-29 15:18:23)

Re: Different Backtest Result using Percent Change Indicator

Hi Popov, I just backtest it with bar closing advance set to 5.
But the result is even more different, please see it
http://s33.postimg.org/k95f8kn4r/SS_29052016_3.jpg
I can't figure out what the problem is. Btw you said that the indicator has no bug, so the prblem lies on the "Bar Closing" point right?
If so, does every strategy that use "Bar Closing" experience this condition?
The number of trades is also different

I’m working on improving my English and trading skill

5

Re: Different Backtest Result using Percent Change Indicator

is there any difference in time between mt4 and fsb?

6 (edited by yonkuro 2016-05-30 02:01:41)

Re: Different Backtest Result using Percent Change Indicator

GD wrote:

is there any difference in time between mt4 and fsb?

Yes you're right the GMT is different. Anyway I tried to test it on various GMT data, and got various results. I never thought this daily strategy could be really time sensitive, although I banned what I think time sensitive logics such as entry & exit hours. I think it'd bad to rely on this kind of strategy, as every broker would experience DST.

However I don't think that GMT difference is the only cause of different result, as Popov said that the orders weren't closed properly.

Anyway, thanks GD.

I’m working on improving my English and trading skill

7 (edited by yonkuro 2016-05-30 03:55:14)

Re: Different Backtest Result using Percent Change Indicator

I noticed that the Oscillator types are really timezone sensitive on Daily TF,while the moving averages show similar results across various timezones (but still different).

I want to make a daily EA that is robust enough to work on various GMT environment.

I’m working on improving my English and trading skill

Re: Different Backtest Result using Percent Change Indicator

as Popov said that the orders weren't closed properly.

I mean MT tester doesn't close the orders properly. It looks like when the backtester enters in a new smaller time frame zone it starts behaving differently. Sometimes there is 10 points difference between the Bar Close and Position close in the MT tester. We cannot do much.

I'll make further tests.

As general hints:
- uses as shorter periods as you can. The best is when the strategy needs maximum 100 bars.
- use Simple MA methods for smoothing instead of Exponential, Weighted...
- use Close price for the indicators.

Following the upper advises you can produce almost 100% equal results in FSB and in MT tester.

9 (edited by yonkuro 2016-05-30 06:53:02)

Re: Different Backtest Result using Percent Change Indicator

Thank you Popov, I'll try it.

I’m working on improving my English and trading skill

Re: Different Backtest Result using Percent Change Indicator

you might want to use a lot more bars for your testing, I usually go back to 2001 for daily to get enough bars

My 'secret' goal is to push EA Studio until I can net 3000 pips per day....

11 (edited by yonkuro 2016-05-30 12:42:22)

Re: Different Backtest Result using Percent Change Indicator

Thanks Dave, I usually use FSB data/Dukascopy to test my strategy. The oldest data is from 2007. By using Dukascopy data I expect a robust and reliable strategy.

However it's not bad idea to download older historical data from metaquotes so I may give it a try.

I’m working on improving my English and trading skill