5 yesrs back testing... what is the difference between 5 years and one year? why do you need such along period? and then live testing for at least 3 months, why not six months?
Curve fitting is a mathematical term? Thank you.
5 minus 1 is a difference of 4 years. At least 3_months so 6_months will qualify also. I'll be waiting for your mathematical method to the poster's question .... thank you.
I was asking a question about the term, I have just started studying a little bit of statistical optimization in R (Self study) so I can follow the ideas and good advise to succeed in analyzing expert advisors.
I agree with raptor, but maybe a lot depends on what trend is the expert going to follow. For the primary trend three/four years may be sufficient. If you are following secondary trend or an intraday trading or scalping less than 3 years can be also fine.
Also very important is Time Frame. there is for example a big difference between a 15minute time frame and an 8H one. I mean that minimum testing period would be much different.
For me it's not just a question of paper trading time, as you can use real account as live testing.
My method is wait the enough time to find a good performance synchronism between backtesting/forward testing/live paper trading testing and small lot size live real trading testing before start to increase real account volume.