I think there's a few problems here. Maybe you should try and work out why some weeks are more significant than others? Non-farm payroll, first week of the month for example. The drawdown is pretty big compared to the profit after 5 years. There's only 147 trades, whats the chances of you losing your balance if you randomize it? I think if you had some reasoning behind why some weeks work better than others then it could be tradeable but if you don't then its not tradeable for me, especially as the drawdown is huge.[/QUOTE
Good point, a monte-carlo test might be beneficial to get a good idea of an acceptable starting balance. What I meant by 'minimum starting account size' was just that and by no meant safest starting account size.
Removing the weekly filter does not make the system less profitable, in fact it makes it more profitable. The reason I introduce it was because the profitable trades seemed very heavy around 2008. By introducing this filter the system gives a more balanced return each year.
