Two constant forecasts give smaller 2024–2025 ZTF g and r errors than the published sine for PKS 2131−021

We test the fixed optical sine from Kiehlmann et al. (2025) on later public ZTF DR24 data for PKS 2131−021. All printed model parameters stay fixed. The early interval ends on 2023 September 9. The test interval starts on 2024 January 1 and has g and r data through 2025 October 21. For equal-weight 20-day means, the sine root mean square errors are 0.799 and 0.744 mag in g and r. Constants from the early means give 0.568 and 0.490 mag. The published constant means also give smaller errors. The result persists with other bin widths, individual object IDs, program 1 data, later date cuts, and a nearby-source calibration correction. Most extra squared error comes from a mean brightness offset. After mean removal, sine residual scatter is slightly smaller than constant residual scatter. An added early-fitted correlated-noise comparison depends strongly on the noise time scale and cross-filter covariance. It does not give a binary probability or global periodicity significance. The i record has only five later bins in 2024 and gives a mixed comparison. The result limits the predictive value of this exact optical point forecast over the tested interval. It does not exclude an underlying periodic component or a black hole binary. The package preserves raw responses, source parameters, provenance, code, numerical checks, tables, figures, and reproduction instructions.

PKS 2131-021 · PKS J2134-0153 · ZTF DR24 · optical variability · forecast test · black hole binary candidate · correlated noise · reproducible astronomy