Hi Roger:

Don't know where the site went, but I'm sure Canie could enlighten us. Maybe her server got whacked up side the head by a rolling blackout .....

I don't think overlapping windows will have any impact on the final outcome of the calculation. Not only EQs will be in multiple windows, but so will the misses, so it's a wash IMHO. Think of it as a virtual predicitor making the same prediction over and over and over again, but each prediction being one minute later. When you summed it all up at the end of say 30 years, you would get the probability.

With regards to your fractional suggestion, it does address the odd size prediction, but your still going to end up with a chunk of catalog as a remainder that you are not addressing. You are still arbitrarily (sp?) applying your test windows to the catalog. I don't like the idea of ending up with two (or more) different answers for a single analysis (the ole 1% -vs- 2% arguement). There should be one answer and one answer only, with an absolute minimum of subjectivity.

I like the idea about quality scoring. Had the same idea myself, except misses would count as 0, as in 0 value. The concept of a negative value for misses is a worthwhile idea and I shall ponder it (after midterms mind you).

I don't have a problem with hard work, so lets keep at it!

Michael