Early sinusoid forecast scores change with the spectral fit for PG 1553+113 in 2015–2026

We tested an early sinusoid forecast for PG 1553+113 with four saved Fermi LAT Light Curve Repository products. All parameters used 83 bins before the 2015 discovery-data end time. The test used 136 later 30-day bins through September 2026. Against each product's early constant mean, later mean squared error fell by 2.83% for fixed-index photon flux and rose by 9.73% for free-index photon flux. The fitted periods were 795.2 and 820.0 days. An exploratory swap of the early fitted periods reversed both photon-flux score signs. Free-index energy flux gave a 13.35% error reduction. All four conditional block intervals included zero improvement. A separate proof showed that a stated published peak rule must accept every window with enough measurements: the selected peak lies inside the acceptance tolerance by construction. All 509 windows passed in a 100-phase stress test, while whole-flux error changed. This audit tests the written rule, not the authors' code. The contribution is a reproducible forecast and method check for a known period candidate. It does not establish or exclude periodicity. The package preserves exact public data, hashes, code, fits, losses, figures, and reproduction instructions.

PG 1553+113 · Fermi LAT · gamma rays · blazars · periodicity · forecast validation · spectral index · public astronomical data