This paper sets out the basis for modelling likely percentages of students at each of the 9 to 1 grades in GCSE maths and GCSE English in 2017. The modelling is based on 2016 outcomes in the previous versions of the specifications. It is based on the underlying distributions of candidate marks (either uniform or raw marks). The paper outlines a number of assumptions involved in modelling. Some assumptions are more straightforward than others.