does the terminal automatically dl new historical data?

New comment
 

i run the ea I was building 2 days ago on tester.. the results were great. then I'm running them now and it looks very bad, its so frustrating.. spent a good amount of time trying to figure out if I had the settings wrong or if I had inadvertently changed something... Has someone come across this before? I guess my question is could MT have downloaded new historical data in the past 2 days?

Furthermore, does anyone have actual historical tick data (one with real spreads) I can use on my tester for this ea?

 

The tester uses the current spread the moment you press start. So one run you might have gotten 2 pip spread and another 3, thus the different results.

The ask line (spread) is not in any mt history and mt history only has bars, no tick data.

If you want to fix the spread do your testing offline. If you want to save a spread value, make a copy of history\ACCT\symbols.sel

When I'm optimizing, I do one variable at a time, one after the other. I only look at the best of that run. I do not look at the actual profit of the run. FWIW

 

Personally, I do all my testing on a disconnected instance of terminal.... In Tools/options/server... put in a dummy proxy server and enable... that will prevent terminal from connecting and you get half a chance of consistency in back test results.

hth

V

 
Viffer:

Personally, I do all my testing on a disconnected instance of terminal




Me too

Viffer:

put in a dummy proxy server and enable... that will prevent terminal from connecting and you get half a chance of consistency in back test results.


What is the advantage of doing this versus just selecting "cancel" and never logging in when the terminal is opened?
 
1005phillip:
What is the advantage of doing this versus just selecting "cancel" and never logging in when the terminal is opened?
I have seen cases where doing that causes the Terminal to 'lose' the account's market properties and revert to some weird 'default' properties. This never happens with a bogus proxy, so that's why I use it. But I guess u can use whatever works for u...
 
WHRoeder:

The tester uses the current spread the moment you press start. So one run you might have gotten 2 pip spread and another 3, thus the different results.

The ask line (spread) is not in any mt history and mt history only has bars, no tick data.

If you want to fix the spread do your testing offline. If you want to save a spread value, make a copy of history\ACCT\symbols.sel

When I'm optimizing, I do one variable at a time, one after the other. I only look at the best of that run. I do not look at the actual profit of the run. FWIW


Thanks that was very helpful.. come to think of it, the one thing I did different was I created an account to demo test the ea.. since I did that the EA just stopped being profitable in tester.. that's probably what screwed it up.
New comment