Strategy Generation: How to best utilise my data?

Ok Title is probably not the best description of the topic

 

If I have 10 years of data, how much of this should I be allowing for my strategy generation, including OOS and how much for other testing i.e. How much should I be keeping up my sleeve so to speak for WFA and Optimisation?

 

e.g. should I be using 80% (8 years) for generation (for both INS and OOS) and the rest (2 Years) up to most current date for the WFA/optimisations and additional testing?

 

How are other using their data that works for them?

 

Regards,

 

Stephen.

 

I use 01.01.2004 ~ 01.01.2008 for my IS, then for OOS i use 01.01.2008 ~ 01.01.2010 when generating,

when retesting i use 01.01.2004 ~ 01.01.2008 for my IS, then for OOS i use 01.01.2008 ~ 01.10.2015 (ALL DATA)

when improving i use 01.01.2004 ~ 01.01.2008 for my IS, then for OOS i use 01.01.2008 ~ 01.01.2010

for optimization i use 01.01.2004 ~ 01.01.2008

then retest it on full data + 500 ~ 1,000 montecarlos

*forgot to mention that i have data from 04.04.2003~01.10.2015, so when i’m retesting in SQ/MT4 i include all the data i have from 04.04.2003 to see the Back-Ward part to see if the strategy was profitable even before it was even built, you can read my suggestion for a BackWard OOS and a 2nd OOS testing parts for SQ4.

hope it helps.

I do everything from 2005 to 2014. 70% IS and 30% OOS.

 

I include 2015 data when I’m testing in MT4.