YES i said that " PDE has a greater magnitude of completeness than NOA" and it is true if you convert PDE and NOA magnitudes on the same scale (i.e Ms) and compare in order to convince yourself. The conversion factor between Ml and Ms is Ml+0.5=Ms for the NOA catalog. For example if you do this you will find that on a common magnitude scale lets say M then PDE will be complete for events with magnitudes greater than lets say Ms=4.5 or even 5 whereas NOA's catalog has been tested
several times and publications show that it is complete for magnitudes greater than Ms=3-3.5 depending on the region of investigation and this beats PDE any day.
You are right about declustering, i have a lot of workr on this regarding NOA's catalog together with Max Wyss in 1999 and after declustering the same years you used 1990-1999 (this time) using Reasenbergs algorithm for crustal earthquakes (that means depths < 50 km> this time not like the last statistic i sent you that concerned depths down to 195 km (that is we were examining more events because of depths and one additional year 1999) we get now 254 M(local)=4 events in 9 years so instead of one event every 7.8 years now we go up to one M=4 event every 12.8 days. Still too close to Romino's time window to be considered as a serious forecast.
Regarding how many "quiet windows of that time length our catalog has is a topic to be discussed upon a succesfull forecast and not at this early stage.
Best regards