Very big difference between backtesting and live trading of EA, any explanation?

Back to topics list To post a new topic, please log in or register
avatar
4
Stephandegroot 2015.02.19 04:30
 

Maybe someone can help me with the following problem. I'm working on a expert advisor now yesterday it was working around the publication of the minutes of FED.

The expert advisor did trade however it traded not profitable, because it closed the position very quick(as per EA). But now I'm back testing and he trade profitable(because is opened the position earlier) 

During the live trading the EA was not trying to open a position, see pictures of the back testing and live trading below. Hope someone can help.

.

Backtesting


Live trading

avatar
1211
deysmacro 2015.02.19 04:34 #
 
Live trading always have setback. Nothing always going smoothly.
avatar
4
Stephandegroot 2015.02.19 04:40 #
 

Yes offcourse I'm aware of that. But the difference is really big in this case I think. And with the live trading the ea didn't want to open a trade till it opened it in the picture, while back testing it opened in way before that.

This difference is normal with back testing? The difference between a 35 pip profit or 10 pip lose looks to me really big?

avatar
1211
deysmacro 2015.02.19 04:43 #
 
You should do a forward test at demo account first before going for real account and work the logic from there.
avatar
Moderator
3404
GumRai 2015.02.19 06:17 #
 

Live trading around news times often have jumps in price

The live EA will suffer because of the jump, whereas the strategy tester will get a price even if it is unavailable in real time. 

avatar
4
Stephandegroot 2015.02.19 06:34 #
 

Ok, thank,

I did some forward tested, and already traded with the ea live where it did work.

The explanation of Gumrai sounds good, the strange thing was that the EA was not trying to buy, however i'm sure there were moments where he could buy.

Thanks for the info, I will continue working on the EA

avatar
12983
WHRoeder 2015.02.19 13:36 #
 
The tester doesn't have gaps. SL/TP are always at set value, no slippage. Did you set a reasonable average spread?
avatar
4
Stephandegroot 2015.02.19 13:39 #
 
In the EA I don't use SL/TP, and I used the spread at current.
avatar
49
gadget 2015.02.23 17:19 #
 

Stephan your question(s) is(are) quite vague.

Testing a system on historic data has limited interest, a bit like unit testing an IT system - keeing aside that the system needs to be limited as well to be testable in MT. That being said, in the tester window you should be able to get the chart where to check what you claim to be wrong, and the good thing of the tester is it will generate all traces if you have put some.


Back to topics list  

To add comments, please log in or register