This is where something more than just searches from history, which is helpful, but...

If you say save the closest 5 (and make programmable) for each search,( say by
time frame) that might help nail one
down, but also you need to compare each of
the five closest to each other as well, just
in some of the hits were off by 1 EQ level
that could knock out 3 misses that could
have been hits, since you were looking for
the wrong collerations. Kind of throwing in
a little std deviation on each level, so that
all hit would count, and not all misses would could either.

I database of quakes to compare to is great once
it's narrowed, but from there, you need to be able to build a new database of quakes that you seem sensitive to, then by comparing each of these
you could start to see some other patterns, or start to see some multiple patterns which sometimes can throw off what would seem like a hit to a miss and you just can't figure out why it wasn't what you thought.

Excuse the rambling, just thinking out load.

George