what constitutes good data for testing?

fundjunkie

Member
Messages
56
Likes
0
Hi,
I'm moving towards undertaking, for the first time, a backtesting project for the Dow and S&P Mini contracts. Can somene tell me, in their opinion, what constitutes good historic data for testing (other than sample size)?

I've seen posts where correspondants claim a data source is good because it is clean. Others claim that raw, unfiltered feeds are good because they haven't been cleaned (can see the logic of that). Still others make reference to the timestamps in a data source?

Maybe someone can improve the signal to noise ratio for me and outline some of the pros and cons in light of the type of testing (generally speaking) I'll be doing.


Thx,
D
 
Top