You're suggesting pre-filling future data based on extrapolating current rules (up to an arbitratily unknown high date) instead of using a constant.

I don't really see how that is a good approach in any capacity.