Tickstory

I am testing my expert advisor working on 1M timeframe (mt4 build 745). Actually its logic uses tick by tick price feed and close/open 1M prices.

I am downloading tick data using Tickstory for 1M timeframe. The strange thing is that I obtain absolutely different results of backtesting if the concerning period of testing is downloaded as a part of a large period of time or individually. For example : I download 1M data from 2001/01/01 to 2011/01/01 and do testing for this period or I download data from 2001/01/01 to 2005/01/01 and then do testing from 2001/01/01 to 2011/01/01 - in these two cases the results are quite different. Is it possible that Open/Close prices of 1M time frame bars c are calculated differently in these two cases ? Thank you for your valuable help in advance.

Hi Trender,

The first thing I would note is that you are back-testing a large period - around 10 years - in one go. There are limitations on how much data MT4 can process and it is most likely you are running into some issues here. You can see here for further information on this: http://www.tickstory.com/forum/viewtopic.php?f=4&t=462.

To answer your question, there should be no difference in the Open and Close prices of the 1-minute bars. The bars are constructed from the time-stamps of each tick that occur within each minute. Once you resolved the issue above, if you are still having discrepancies I would suggest you analyse a few trades and compare the timing of the trades. Bear in mind that if your EA is somehow dependent on the starting date (for eg. it only takes a trade if the last trade was successful), then will you obviously get different results depending on the range you select.

Hope this helps.

Thanks for your comments and indices !
I have noticed also that the size of M1 history file, eur/usd from 01.01.2010 to 01.01.2011 is abnormally small ( 50 %) with respect to sizes of similar ones for 01.01.2011 - 01.01.2012 and 01.01.2012 - 01.01.2013. Is it a market info which was too particular this year having thus smaller number of ticks, or some gaps were present in the downloaded file ?

Hi Trender,

We are not aware of anything aside from market conditions which could affect this. Generally speaking, the Dukascopy data across most pairs for the past few years is quite comprehensive.

Regards.