Help setting up ASX Index

[danny_pipParticipant January 26, 2015 at 11:46 am Post count: 26]

#20478

Hi,

I’ve added the screenshots and settings of csv2fxt I’ve used.

Thx for the efforts !

danny

[imported_tickstoryParticipant January 26, 2015 at 12:42 pm Post count: 1479]

#20479

Hi Danny,

It appears both your screenshots are produced from CSV2FXT as it indicates a 99% modelling quality rather than one indicating Tickstory’s “99.9%” modelling quality.
What you need to do is produce the FXT files using CSV2FXT and then Tickstory and compare the FXT Header information to see if there are any differences.

Hope this helps.

[danny_pipParticipant January 26, 2015 at 12:53 pm Post count: 26]

#20480

I don’t understand exactly what you mean;

I did an export to a CSV file with SQ TickData and converted these with the script CSV2FXT. Of this I’ve made a screenshot using your ‘MT4 Data editor’.
Next I did an export to a CSV file with TickLite using the format you gave me. This CSV file I’ve converted with CSV2FXT. Of this I’ve made again a screenshot using your ‘MT4 Data editor.

If you like I can give you access with ‘teamviewer’ to my computer.
Or with skype to make things more clear.

[imported_tickstoryParticipant January 26, 2015 at 1:27 pm Post count: 1479]

#20481

Hi Danny,

As I understand you are seeing differences with CSV2FXT-produced data versus Tickstory’s “Export to MT4” function. Tickstory’s “Export To MT4” function does not require any export to CSV but instead directly produces an FXT with the parameters you give it. Birt’s CSV2FXT on the other hand produces it’s own FXT file from a CSV file and takes parameters from the .set file you provided. From what you have described, you are currently comparing two FXT files which CSV2FXT has generated, meaning that both FXT headers will invariably be the same.

To compare, you therefore need to export the data once using CSV2FXT (using SQL Downloader or Tickstory CSV data) and once again using Tickstory’s “Export to MT4” feature. In both cases, take a screenshot of the MT4 FXT Headers so you can compare to see if there are any differences.
The idea is to compare the FXT header data between CSV2FXT and Tickstory as it is most probable that there are differences in this header that are causing your discrepancy.

Hope this clarifies.

[danny_pipParticipant January 27, 2015 at 6:19 pm Post count: 26]

#20482

Here it is.
The results with every generated FXT file are so different that I don’t really know at all what is correct or not.
I’ll hope you can figure it out.
There ‘s also a big difference in lotsizes that I have to use.

Regards,

Danny

[imported_tickstoryParticipant January 27, 2015 at 11:55 pm Post count: 1479]

#20483

Hi Danny,

Comparing the FXT headers of your CSV2FXT vs Tickstory export, there are the following differences:

Spread=2 (in Tickstory you have exported a spread of 3).
Digits=0 (in Tickstory you have it set to 1)
Margin Init=0 (in Tickstory you have a value of 10000)
GTC Pending=Yes (In Tickstory you have it unchecked)

Could you please try re-export to match the changes described above? You will need to manually change the respective Metatrader Info fields and then “Save as default”.
Then try your back-test again and see if starts to match up with what you expect.

Regards.

[danny_pipParticipant January 28, 2015 at 9:21 pm Post count: 26]

#20484

Hi,

I’ve done what you’ve asked; see attachment.

Regards,

Danny

[imported_tickstoryParticipant January 29, 2015 at 1:09 am Post count: 1479]

#20485

Hi Danny,

Thanks for the information – it looks like you have got the correct base-line for the Metatrader Info/FXT header information which is a good start. From here you need to identify any differences in your back-testing setup. For one, I can see that you have different export dates. You will note that in each back-test you have a different date for your first trade. Depending on how your EA is designed, it could mean dramatically different results. For example, if you happen to be in the midst of a trade in one back-test vs the other, it may mean you miss out on a number of profitable trades that the other is taking. Or if you happen to be relying on a certain time/date/indicator/money management, this condition could be different in each case (also indicating that your EA could be vulnerable to different start times). To eliminate some of these variables, make sure you export the same date range for both back-tests. Of course, if you see any discrepancies with the data that Tickstory is producing, we will investigate further.

Hope this helps.

[danny_pipParticipant January 29, 2015 at 8:50 am Post count: 26]

#20486

Hi,

the beginning date does not matter in this case; because the only valid tick data start form 23/02/2013.
So if I take 26 jan 2016 or 27 jan 2013 it does not matter.

the problem has to be somewhere else.

Thx for your reply

[imported_tickstoryParticipant January 29, 2015 at 9:44 am Post count: 1479]

#20487

Hi Danny,

The only thing I can suggest is to take the first trade as an example and see why the EA takes it in one case but not the other. Is it due to an indicator? In the most base case it would be worthwhile checking the two sets of data have the same candles and/or indicators at that given period. If there is something notable we might be able to assist further.

Regards.

[danny_pipParticipant February 2, 2015 at 8:44 am Post count: 26]

#20488

Hi,

I’ve done an export of the last month for every system.
Now I’m getting similar results but still different.
I’ve also noticed that SQ Tickdata Downloader did not exported all files of 2013; Some were missing.
To be sure that I’m working with good tickdata, I’ve took the last month; what’s obvious is that the sizes of the FXT files are different.
This leads also to different backtests.
Now the question is: which are the best ?

Here’s the link where you can download the results (the file is too big to put in attachment).

http://users.telenet.be/karaoke/results.zip

Thx for all efforts again !

Danny

[imported_tickstoryParticipant February 3, 2015 at 12:09 am Post count: 1479]

#20489

Hi Danny,

All I can suggest at this point is to try export a small sample of data to analyse what the differences are. In the next major upgrade of Tickstory we will be offering a viewer that can browse the MT4 FXT files which may make it easier to see if there are any differences in the two data sets you are exporting. Until then, it may be necessary to just manually export them (to Excel for example) and see if its the actual bar formations that are causing your differences or some specific setting (such as “Filter weekends”). This is of course assuming that your Metatrader Info/FXT Headers are now the same.

Do note that as long as you see no errors appear in the Tickstory “Log” window during your export/download process you can expect the data is complete and matches what Dukascopy have on their servers.

Regards.

[danny_pipParticipant April 9, 2015 at 4:07 pm Post count: 26]

#20490

Hi,
The broker FXCM has recently added 2 digits to the AUS200 index; instead of 5946 it’s now 5946.00.

Some brokers like Sensus uses no extra decimals; Pepperstone uses 1 decimal; I’ve noticed that this causes problems again in exporting and optimising.

Backtesting does not work anymore.

Hope you can get me out again !

Grts,

Danny

[imported_tickstoryParticipant April 12, 2015 at 4:48 am Post count: 1479]

#20491

Hi Danny,

Could you please detail the issues you are having? Are you getting an error? Is the price coming out correctly when back-testing?

Thanks.

[Author]

Posts