Modeling quality changes based on timeframe

[shani]

Hello,

I exported EUR/USD tick data to MT4 and lunched MT4 for back-testing through Tickstory.
When I run a back-test on a 1H chart I get 99.9% modeling quality, but when I run the same back-test on the daily chart I get modeling quality N/A.

What can be the problem?

Thanks

[tickstory]

Hi Shani,

Make sure you have exported FXT files for every timeframe. You can double-check within the MT4 Data Folder and make sure that all the FXT files exist for the other time-frames and should be the same size as EURUSD60_0.fxt (i.e. the 60 minute/1 hr timeframe).

Hope this helps.

[shani]

I exported to MT4 again and got 99.9% in all time frames for a week or so.
Yesterday I ran a backtest on 4h EUR/USD and got modeling quality N/A again.
I've checked the files under MetaQuotes\Terminal\C4F6B9F5EDF8FD79C0DA897541DC313E\tester\history
and the file EURUSD240_0.fxt is a lot smaller than the other files and the filed "date modified" shows yesterday's date even though I haven't exported data or changed data files. All other files in this folder shows under "date modified" the date I last exported data to MT4 (24/10/2014).

What can cause this change in the file?

[tickstory]

Hi Shani,

If you are getting N/A modelling quality, it usually indicates that you
1) Never launched MT4 via Tickstory and/or
2) Didn't generate the FXT for that particular time-frame.

You can trouble-shoot this by checking that the relevant FXT file exists immediately after exporting it via Tickstory. Ensure it is the appropriate size & is marked as 'read only'. Then, launch MT4 via Tickstory and attempt your back-test. You should hopefully see no changes to the FXT file and get 99.9% modelling quality on your back-test.

Hope this helps.

[RV13]

What effectively happens when MT4 is launched via tickstory?

[tickstory]

Hi RV13, the launcher prevents MT4 from overwriting the FXT files that Tickstory generates. Without it, MT4 will attempt to generate the FXT files again but without tick data.

Regards.

[shani]

Hi,

I have another issue when exporting data to MT4:

In the past, I've exported tick data from 2012.01.01 22:00 - 2014.11.05 23:5 and ran all my back testing on these dates: 2012.08.07 - 2014.08.07

Now I wanted to expend the testing period, so I've exported tick data from the last 5 years (2009.12.22 00:00 - 2014.12.21 23:59).
Then I ran the same back testing test on the same dates: 2012.08.07 - 2014.08.07 as before and got completely different results (all the statistics are different :profit, number of trades and so on).

On the original report this was the info I got:
Period: 4 Hours (H4) 2012.01.01 22:00 - 2014.11.05 23:59 (2012.08.07 - 2014.08.07)
Bars in test:4595 Ticks modelled:55815300 Modelling quality:99.90%

Now I get this info:
Period 4 Hours (H4) 2009.12.22 00:00 - 2014.12.21 23:59 (2012.08.07 - 2014.08.07)
Bars in test 8073 Ticks modelled 76695831 Modelling quality 99.90%


Why is the number of bars different for the same testing period and time frame?
Why do I get different statistics and what should I do?

[shani]

I've tried to uninstall Tickstory and MT4, deleted the tick data folder and started from scratch...
But I still get almost double bars count for the same test period and time frame
What should I do to?

[tickstory]

Hi Shani,

Make sure your FXT files are less than 4GB in size as there are limitations within MT4 that could be causing your issue.

See here for more:

http://www.tickstory.com/forum/viewtopic.php?f=4&t=462

Regards.