Thanks for kind words, footon. 1M is fastest to find out if strategy is good or not. If I would use 1d they I would have to wait 2 years In August and September I have some 5m strats, but they are too slow.
Do you think that using 15m would be better or less random?
I used 30k bars for generating strategies and also OOS 30%.
Settings below (sometimes I used perm s/l with 50 or 60 pips)
http://screencast.com/t/et8aIUjxDH64
and
http://screencast.com/t/FSXuxumgXRb3
My next move would be to change the losers faster and not hope they will get better as you can see on the chart. I plan to gather few of such strategies and run some micro real account.
Winning strategy gives 6% a month which is not bad. If I would have at least 2 such strategies working for 1 year, it would be good.