There are already som any coeficient influencing this (price, capacity, some believe survey, recent results, maybe a rate between different kinds of seats,...) why to add another no clear thing like cups? Do it with minimum influencing things possible to get bet results. It can be always recalculated with other types of games invloved but nobody confirmed that attendance for cup games is claculated the same as for league games (unless i overlooked somthing).