We were presented with a new author compensation scheme by Fawce here:
Royalty = (weight of algorithm in signal combination) * (total net profit of the combination)
Presumably, Q is using something along the lines of this architecture to combine the signals:
It is worth noting that there is not a 1:1 correspondence between the weight assigned to an algorithm in signal combination and the contribution of that algorithm to the overall fund return (in fact, there may be forms of alpha combination for which the concept of a weight does not apply, since one does not need to do a simple weighted linear combination of factors). There's a certain amount of voodoo here, in my opinion, that needs to be clarified.
Does anyone know how the weights are determined?