I would be curious regarding which quake it is that is least probable - the Ml 3.4
near Big Bear or the smaller events in Northern California. As you stated, the
larger events will always be the least probable though. The real problem is determining
how to match forecasts with events. If the Big Bear earthquake lies near to
a forecast the question is not how common the Big Bear event was or how
often a forecast will be right, but what is the probability that the Big Bear earthquake
would occur given that prediction X had been made. In other words it is a
Bayseian probability question.
In this case the question is: what is the probability that an event within the magnitude
window forecast would occur within a 7-day span and within 120 km of the forecast
epicenter. Simple probability of occurrence of an earthquake should not be the
important factor, but what the probability that the forecast would have been correct
given the expanded limits which would include the event outside the forecast
boundaries.
For example, if I predict an event of Ml 2-4 within 40 km of 33N 117W and one
occurs in the time window at 34.3N 116.8W of Ml 3.4, the probability that must
be addressed is : what is the probability that a Ml2-4 event would have occurred
in the region within 130 km of 33N 117W (which I suspect is pretty high) in the
7-day window.