Correction: it will go up given an ignorably small jerk on the line at the lowest point. Not the best example, right... What I was trying to say is that one can trade rotational momentum to linear (and vice versa) as long as the kinetic energy of the system stays the same.