Late M5 event excess remained after one-year isolation on both sides of 82 deep M6.5 to below 8 earthquakes in 2001-2024

We tested whether a late event excess after deep earthquakes depends on an isolation rule that uses only past events. We reused a fixed USGS ComCat snapshot. Centers have depth greater than 300 km and preferred magnitude 6.5 to below 8 in 2001-2024. The main rule excludes centers with an equal or larger event within 200 km horizontally in either the previous or next 365 days. For 82 retained centers, M5 or larger neighbors within a depth-aware 200 km distance gave 29 after counts and 12 before counts on days 7-30. The excess was 0.00901 events per center per day, with a conditional 95% three-year block interval of 0.00555 to 0.01329. The past-only rule retained 95 centers and gave 0.00778. The difference between these estimates was uncertain. Future-only isolation also gave a positive interval. However, the both-side interval included zero with a 30-day isolation window or a 300 km neighbor radius. The result survives reversal of the isolation direction at the tested one-year horizon. It does not remove selection effects, identify caused aftershocks, or establish a physical recovery time. Sparse counts, regional dependence, mixed magnitude types and catalog revision limit inference. The package gives new code, aggregate evidence, audits, a figure and exact external input retrieval instructions.

deep earthquakes · mainshock selection · temporal asymmetry · late event excess · ComCat · negative control