On the Curses of Future and History in Future-dependent Value Functions for OPE