mt5 strategy tester ticks - page 2
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
A title is just a couple lines a text to name an article so it can be found by users (like little ole' me) when surfin' the site. Yes, the article's title does give strong indication of its subject's background but I decided to read its content in order to receive a detailed explanation. Yep, I can't argue with the article's title "Algorithm of Ticks' Generation" but I feel it doesn't really help me when I haven't read the article's content as confirmation for my question (not too hasty now WhooDoo, right? Hahahaha!)
Hi WhooDoo22,
They are not accurate.
This would be a quite important issue for me. I try to develop fast strategies (trading on 5 minutes chart, deals last for minutes only). I got good results in the tester optimalisation but in real time the results were much different. And different means not only profit or loss, means numbers of deals. Obviously, when based on the test you expect few trades a day but on real time demo account dozens happen you suspect something is wrong with the tester.
Therefore I started look after this problem. I wrote an EA which does not trade, only records ticks into file. This gave the real life data (it runs on VPS, so it is reliably recording everything). I created a modified version also which prints out each tick data from the tester. I extracted this piece from the log. So, I had both data and could compare. And the surprise came.
Hi WhooDoo22,
They are not accurate.
This would be a quite important issue for me. I try to develop fast strategies (trading on 5 minutes chart, deals last for minutes only). I got good results in the tester optimalisation but in real time the results were much different. And different means not only profit or loss, means numbers of deals. Obviously, when based on the test you expect few trades a day but on real time demo account dozens happen you suspect something is wrong with the tester.
Therefore I started look after this problem. I wrote an EA which does not trade, only records ticks into file. This gave the real life data (it runs on VPS, so it is reliably recording everything). I created a modified version also which prints out each tick data from the tester. I extracted this piece from the log. So, I had both data and could compare. And the surprise came.
Actually, the tester data is more. I expected that the tester data is less because of simplification explained in this article https://www.mql5.com/en/articles/75, but it is not true. Just to reiterate with simple words to make the point clear: in the strategy tester, more ticks are generated for the same time period (for example 1 minute) than there were in real life. Moreover, volumes are totally different showed by the built in indicators than recorded.
Ps:
The problem with the difference in number of ticks tester vs real life is not transparent because the main candle data (open, close, high, low) agree. Without recording real life data and compare to tester is not possible to recognise it.
Hello NyemaSanya,
Thank you kindly for your thoroughly explained point of view.
A solution could possibly be if whoever has power to modify the MQL5 tester code modifies it so as to read real tick data from a history folder contained within the MQL5 home directory folder like that for MQL4.
I always have a good laugh when running strategies within MQL4 tester in every tick mode (90% accuracy) then switch on "nit-picky" mode (99% accuracy) and read "the writing on the wall" when entry/exit signals show entirely different results.
Thank you
You haven't read the article ? the same article I gave you a link to on the 31st Jan via PM.
Oops, wrong word.
"I haven't read the article's content" to "I hadn't read the article's content".
Yes, I believe I read this article you provided me.
Thank you
How do you determine if you have missed a tick or several ticks ? and what do you do if you see you have missed a tick or several ticks ?
I'm guessing you're speaking with NyemaSanya, yes? All I know about the MQL5 tester is it runs fake ticks (seemingly more accurate than MQL4 tester) and that's about it, sir.
Thank you
How do you determine if you have missed a tick or several ticks ? and what do you do if you see you have missed a tick or several ticks ?
Hi RaptorUK
The difference is not one or two missing ticks. I give you an example: 2013. March 7, from 2.00 till 10.00. Number of ticks in real life 27 878, in tester 49 676.
Hi RaptorUK
The difference is not one or two missing ticks. I give you an example: 2013. March 7, from 2.00 till 10.00. Number of ticks in real life 27 878, in tester 49 676.
Hello NyemaSanya,
Thank you kindly for your thoroughly explained point of view.
A solution could possibly be if whoever had power to modify the MQL5 tester code modified it so as to read real tick data from a history folder contained within the MQL5 home directory folder like that for MQL4.
I always have a good laugh when running strategies within MQL4 tester in every tick mode (90% accuracy) then switch on "nit-picky" mode (99% accuracy) and read "the writing on the wall" when entry/exit signals show entirely different results.
Thank you
How do you determine if you have missed a tick or several ticks ? and what do you do if you see you have missed a tick or several ticks ?
So how many ticks do you typically miss ? if you don't check then you wouldn't know if you are in fact missing 50% of the ticks ?
Sorry RaptorUK,
I do not really understand your question. It looks like it is not clear for you that the more ticks are in the tester, not in real life. Therefore I do not miss ticks from the tester, I would like to throw out the extra ones from them (because I trust in that my real time data recorded on VPS is right and complete). In the example above, 49676-27878=21798 additional ticks are generated by the tester. (it is EURUSD data on Alpari broker, I have forgotten to mention it, although it probably does not matter).