I'd like to repeat - this EA is not the real expert, it's just test for Indratime's idea. He got some really good ideas about price movements, about general direction and small steps back before going ahead.
I'd done EA to test his ideas on long history as it's hard to do it manually. These two levels - 10 and 20 points - after midnight were really good.
The system is profitable, but not so profitable as it would deserve real market test. It can stay around 0 or even with some losses too long.