On the accuracy of spherical harmonics and orbital predictions