Early largest-event magnitude gave no clear count forecast gain after 323 shallow M6.5+ earthquakes in 2013-2024

We tested whether the largest nearby event magnitude in days 1-7 improves a forecast of M5.5+ counts in days 7-30 after early counts are known. We used a fixed ComCat snapshot, 363 training centers in 2001-2012 and 323 test centers in 2013-2024. Centers have M6.5 or more and depth at most 70 km, with no equal or larger event in the preceding 30 days within 300 km. The primary neighbor radius is 200 km. Adding early maximum magnitude to a Poisson mean model with early count, center magnitude and depth reduced mean test deviance by 0.00409, or 0.26%. The conditional 95% three-year block interval was -0.01673 to +0.02820. Four planned sensitivity intervals also included zero. Thus this test gives no clear support for the added term. Sparse early events, revised catalog records, mixed magnitude scales, shared windows, simple model form and twelve test years limit inference. This is not evidence against secondary triggering or a live forecast validation. The package gives code, provenance, aggregate evidence, a figure and an independent audit. External event records are not bundled.

earthquake · ComCat · forecast · early maximum magnitude · aftershock count · null result