Testing EAs...some issues

Hello all, I’m a fairly newbie (gladtomeetya!!),

I’ve got a demo acct on MBTrading using their version of MT4 (modified for an ECN)…and have written a few peanut EAs, but have one that works. Except sometimes.

I’ve tested an EA I wrote with the same compiled code, same external inputs…using data from Jan to June this year…exactly the same date range selected. One day it’s up $2000. Five days later I try it, and it’s down $700. Today it’s up $1200. I run it using use all tick data (“Every tick” setting)…the best setting available.

What gives? I understand Alperi has two years of historical data…is that more reliable? I don’t want to go back too far…things change, right?

When I run the test, I assume I get the tick data from them every time, correct? Perhaps sometimes I get all of it…other (busy) times perhaps I don’t?

Do any of you get huge variances in profits when you test EAs?

-Robert