Hi Roger:

I suppose running your declustering program might give a better estimate of expected probability...presuming that "past performance is an indicator of future events".

Also, care to expand on this:
"This assumes a uniform distribution which we know isn't true..."

$.02