Current stats don't allow too much remixing of the output, need to generate a simpler data structure so that generating the stats can be more easily mixed and matched.