I've talked at some length before about the question of fitting forecast weights, the weights you use to allocate risk amongst different signals used to trade a particular instrument. Generally I've concluded that there isn't much point wasting time on this, for example consider my previous post on the subject here.
