I would also be very interested in seeing this applied to predictions at t + delta as well. Will probable attempt it over the Christmas break