Is the duration of the testing period important? I would think so. I'd rather have a successful system compiled over 12 years of data than one which had many selections over a period of 6 months.

And how far back did bazza's data go darkydog2002?