Topic: Different results btw FSB backtest and MT4 backtest?

Hi there I am using uploaded data from my FXDD demo account to test various strategies on FSB.  I created a donchian channel breakout strategy on FSB using my FXDD data which looked very promising and tested well on MonteCarlo.  However after turning the strategy into an EA and back testing it over the same period on MT4 I am getting completely different results (much much worse).

If FSB is using the same data why would the results on the MT4 back test not be the same?  I can understand a bit of difference due to varying spreads, slippage, etc. but the results are night and day.  It is somewhat troubling because it calls into question all of the back test results on FSB for various strategies as I am creating them.

Any feedback would be great.

Thanks in advance.
Serge

2 (edited by yonkuro 2016-05-18 01:41:53)

Re: Different results btw FSB backtest and MT4 backtest?

sergiel wrote:

Hi there I am using uploaded data from my FXDD demo account to test various strategies on FSB.  I created a donchian channel breakout strategy on FSB using my FXDD data which looked very promising and tested well on MonteCarlo.  However after turning the strategy into an EA and back testing it over the same period on MT4 I am getting completely different results (much much worse).

If FSB is using the same data why would the results on the MT4 back test not be the same?  I can understand a bit of difference due to varying spreads, slippage, etc. but the results are night and day.  It is somewhat troubling because it calls into question all of the back test results on FSB for various strategies as I am creating them.

Any feedback would be great.

Thanks in advance.
Serge

If it's about Donchian Channel Indicator, please read this post http://forexsb.com/forum/post/35701/#p35701

Re: Different results btw FSB backtest and MT4 backtest?

Thanks I have loaded the new indicator into FSB and will run the Generator again and see what happens.

Thanks very much!
Serge

Re: Different results btw FSB backtest and MT4 backtest?

Re generated the strategy and EA with the new Donchian FSB indicator and FSB back test results still completely different then FXDD MT4 backtest for same period.

Any help would be great, thanks.

sergiel wrote:

Thanks I have loaded the new indicator into FSB and will run the Generator again and see what happens.

Thanks very much!
Serge

Re: Different results btw FSB backtest and MT4 backtest?

Did you get different result on MT4 after using the new Donchian Channel indicator?

Re: Different results btw FSB backtest and MT4 backtest?

Serge, you have to post the strategy in order to be able to test it and to find the problem.

Re: Different results btw FSB backtest and MT4 backtest?

Hi there, I have posted 3 images.  The one labeled FSB Panel is obviously shows the backtest for the 15M GbpUsd Donchian breakout strategy using the FXDD uploaded data.

The second image shows the date range selected in my MT4 FXDD demo EA strategy tester as matching the date range in FSB.

The third image shows the backtest graph in my FXDD MT4 demo account for the exact same period starting with $1000 demo account.

I had to post the results graph before the back test was 100% complete as I had to leave for work, but it was about 95% done and in complete draw down so you get the point.

Any help figuring out why the FXDD MT4 backtest is completely different than the FSB back test for the same period using the same data would be much appreciated.

Thanks in advance


http://s32.postimg.org/x39zqlr0h/2016_05_18_FSB_Panel.jpg


http://s32.postimg.org/fl61wb11d/2016_05_18_0835_MT4_Bactest_Settings.jpg


http://s32.postimg.org/ad9ziemf5/2016_05_18_MT4_EA_Backtest_graph.jpg

Re: Different results btw FSB backtest and MT4 backtest?

You use "time" indicators - Entry Time and Exit Hour. When you do so, it is very important to use the same time zone.  Are the data is absolutely the same?

Another issue I see is that the Entry Time indicator has parameters as 10:19 and 14:04. These numbers are not divisible by 15 - your strategy period.
It should work, but it is not sound very professional. You can try with entry from 10:15 to 14:00 for example.

I suggest you to test a short period of time. Let say 1 week. Compare one long entry and one short entry in MT tester and in FSB chart for the same date. It must be no difference in the bar data and the trades. If there is a difference, we will see the exact reason. It must be something like:

- different bar data
- different Donchian Channel value
- different entry bar
- different exit bar

With that info it will be very easy to find and correct the problem.

Re: Different results btw FSB backtest and MT4 backtest?

Hi thank you for your assistance thus far.  Ok I have changed the FSB strategy entry and exit times to the following:

From hour: 10
From min: 30
Until hour: 14
Until min: 0

Exit Hour: 19

I have attached the image for the new FSB strategy settings and back test result.  Net balance $8,942.37 starting with an account of $1,000 over the period of April 29, 2014 to May 6, 2016. 

I have re-saved the new EA based on this strategy uploaded it to my FXDD MT4 demo account and ran a back test in FXDD for a 1 week period only from January 1, 2015 to January 7, 2015.

I have posted the following charts:

1) Jan 1 - Jan 7 2015 FSB Journal;
2) Jan 1 - Jan 7 2015 MT4 back test results;
3) Jan 1 - Jan 7 2015 FSB chart;
4) Jan 1 - Jan 7 2015 MT4 back test chart;
5) Jan 2, 2015 FSB chart magnified;
6) Jan 2, 2015 MT4 back test chart magnified.

I am not sure what to look for to be honest but I can say right away the entry times differ in the FSB back test versus the MT4 back test. 

If part of the issue is a time zone issue how would I address that?  Wouldn't FSB automatically detect which time zone is being used by FXDD if it is using FXDD data? 

Thanks in advance as any assistance would be greatly appreciated.

http://s32.postimg.org/ueompfsn5/FSB_Donchian_Strategy_Settings.jpg



http://s32.postimg.org/pphj36c3l/Jan_1_Jan_7_FSB_Journal.jpg


http://s32.postimg.org/9iy7jto01/Jan_1_2015_Jan_7_2015_MT4_Results.jpg


http://s32.postimg.org/q3jbm5d4x/Jan_1_2015_Jan_7_2015_FSB_Chart.jpg

http://s32.postimg.org/58fetvagx/Jan_1_2015_Jan_7_2015_MT4_Backtest_Chart.jpg


http://s32.postimg.org/omu0ys62p/Jan_2_2015_FSB_Chart_Magnified.jpg

http://s32.postimg.org/ymykpjd1d/Jan_2_2015_MT4_Backtest_Chart_Magnified.jpg

Re: Different results btw FSB backtest and MT4 backtest?

It is very simple.

Data is different!!!!


http://s32.postimg.org/tjiuzfhe9/screenshot_1640.jpg

11 (edited by sergiel 2016-05-19 13:36:46)

Re: Different results btw FSB backtest and MT4 backtest?

Ok, why or how is the data different if I have uploaded and am using the FXDD data in FSB?  And how do I rectify?

Thanks in advance smile

Serge

Re: Different results btw FSB backtest and MT4 backtest?

The most safe way is to export data directly from the chart. You can use the File -> Save As from the MetaTrader menu or to press Ctrl + S key. The other methods also work, but require more caution.

Be sure that the servers are same when import from HST files. Never use the MetaTrader's History Center -> Download button. It replaces your broker's data with data from MetaQuotes.

Re: Different results btw FSB backtest and MT4 backtest?

Thanks again for your assistance as it is much appreciated.

However, I am quite confused now lol.  I imported data from my FXDD MT4, converted to csv files and uploaded to FSB closely following the directions in the FSB tutorial:  http://forexsb.com/wiki/fsbpro_guide/mt4_hst_import_help


http://s32.postimg.org/dp45z9qgx/FSB_HST_Download_Import.jpg

In your last response you said:  "Never use the MetaTrader's History Center -> Download button. It replaces your broker's data with data from MetaQuotes."

This confuses me because isn't that exactly what it says to do in the FSB tutorial for downloading the HST MT4 Data before converting it to csv files and uploading to FSB?

Really confused now.  Further explanation would be most appreciated.

Thanks in advance.
Serge

Re: Different results btw FSB backtest and MT4 backtest?

When you use 'History Center', it downloads from Metaquotes server which is not your broker.

Using the technique of the chart and control S, you are getting the data that your broker is using.

Re: Different results btw FSB backtest and MT4 backtest?

Hi Blaiser, thanks. Yes I realize that now after Popov's post but I guess my question is this: Does the tutorial need to be revised because it seems to clearly conflict with what you guys are telling me now?

Second if I have already used the MT4 history download button as per the tutorial instructions have I now permanently replaced the historic broker data with Metaquotes data?  Or can I simply reboot my FXDD terminal and the historic broker data will be available again?

I have now used the simple - File -Save As function directly from each chart time frame to save the csv data, and then copied the new csv data to the FSB data folder.

Thank you in advance for anyone who takes the time to respond, much appreciated. smile

Serge

Re: Different results btw FSB backtest and MT4 backtest?

To my understanding the downloaded data stays and later on only new data gets added, so the MQ data is there. You should delete hst files, then open the terminal, scroll the charts to have only broker's data stored. I would do this this way. There can be a better option smile

Re: Different results btw FSB backtest and MT4 backtest?

Serge, you are right. I have to fix the article.
I used to download data from the MT History Center, but it was before knowing the fact that it replaces the broker's history.

You can fix it in the way Footon propose - delete your history and download data from the chart.
It is easier when you maximize the screen, set bar chart and zoom-out to minimum resolution and hold the Home button until the chart stops loading older data. It is also faster when you switch off the chart auto-scrolling.
This is the way I'm using now. It takes about 20 minutes to download all periods for 8-10 symbols.

Re: Different results btw FSB backtest and MT4 backtest?

Ok thanks Popov.  I have gone back using the technique you described and downloaded the broker data.  However, the data does not go back very far at all now (only 9 months on 30 Min Chart).  I am hoping it is just because it is the weekend and the market is closed.

Thanks
Serge

Re: Different results btw FSB backtest and MT4 backtest?

Try setting big numbers in the Chart Options:

http://s32.postimg.org/9z4sg0xqp/screenshot_1670.jpg

Re: Different results btw FSB backtest and MT4 backtest?

sergiel wrote:

Ok thanks Popov.  I have gone back using the technique you described and downloaded the broker data.  However, the data does not go back very far at all now (only 9 months on 30 Min Chart).  I am hoping it is just because it is the weekend and the market is closed.

Thanks
Serge

Not many brokers offer data which goes way back, usually it's 20k bars at most...