I see what you're saying. One of the limitations of this system is that because it is so resource-hungry, it's impractical to do a 10-year backtest on all likely pair candidates, at least not until I build a supercomputing center someday :)
As for funding it with a live account, the nature of this system also requires a rather large account for adequate risk management. I wish I had a few million dollars laying around, but that's unfortunately not the case. I could fund it with a smaller account, but I think I'll need at least 100k minimum. I might be able to make it work with even smaller accounts, but the system won't be deployed as intended, the risks might be higher, and risks would not be spread across as many instruments as I'd like.
I suppose I'll just keep forward testing for a few months, and if the results continue to be encouraging, I'll attempt to raise enough money to trade this in a managed account..
By the way, i'm currently broadcasting live video of this system at work:http://www.livestream.com/gatornuke