We analyze a loss cone anisotropy observed by a ground‐based muon hodoscope at Mt. Norikura in Japan for 7 hours preceding the arrival of an interplanetary shock at Earth on October 28, 2003. Best fitting a model to the observed anisotropy suggests that the loss cone in this event has a rather broad pitch‐angle distribution with a half‐width about 50° from the IMF. According to numerical simulations of high‐energy particle transport across the shock, this implies that the shock is a “quasi‐parallel” shock in which the angle between the magnetic field and the shock normal is only 6°. It is also suggested that the lead‐time of this precursor is almost independent of the rigidity and about 4 hour at both 30 GV for muon detectors and 10 GV for neutron monitors.