The prediction must be put into context. If there has been a relative quiet period
in the region during the previous X months, then a prediction of an event in that
region is more significant than one in an area which has been active and is
continuing active at the time of the prediction. Predicting an earthquake during
a swarm is a lot different from predicting an earthquake out of the blue. In this
case there was a swarm going on in the area when the prediction was made - the
seismicity in the period immediately preceding the event is the correct choice to
put the prediction in context. This is, of course, an arbitrary choice and could be
argued, but that is my opinion.