Sample size

1 replies

Enric

Customer, bbp_participant, community, sq-ultimate, 114 replies.

Visit profile

5 years ago #233677

Hi,

I’m wondering how long should be the period to test a new strategy IS+OOS.

Of course to have statistical relevance the testing period should be long enough.

But on the other side, having historical data on Forex since 2003 I don’t think it’s too smart using such an old data because the market probably will have suffered several changes. I mean; who cares to find an excelent strategy in 2006, probably this strategy will rejected in 2016. So why should I look for a tradeable strategy using 2006 historical data?

KR

Enric

0

Ryan

Customer, bbp_participant, community, 0 replies.

Visit profile

5 years ago #233687

Finding a strategy that has done all throughout the years (10 years+) is best IMHO.  My view on this is if it has done well for 10 years, why would it fail when I start it?

Once I find one that has a good 10 year equity curve, I then run it thru the gauntlet of tests.  Usually it fails so it will take some time to find a good one.

Ryan Guderian

0

Viewing 1 replies (of 1 total)