Fixed early sine fits give larger errors than a fixed mean in 2024–2025 ZTF data for AT 2021hdr

AT 2021hdr showed repeated optical changes in an earlier study. We test whether a fixed sine curve from that state predicts later public ZTF DR24 data better than a fixed early mean. The early range is 59500 <= MJD < 60492. The later range starts on 1 July 2024 and ends in October 2025. After quality cuts, the later sample has 126 g and 128 r measurements, in 111 band bins of 7 days. A common early period fitted between 60 and 120 days is 98.21 days. Its later combined root mean square error is 0.1116 mag. The fixed-mean error is 0.0990 mag. The ratio is 1.127, with a 95% paired 30 day block interval of 0.952 to 1.308. The interval includes equal performance. A fixed 97.7 day model gives a ratio of 1.131. Tests with 3 and 14 day bins and an early source-shape correction also give larger measured sine errors. The correction leaves substantial transfer error against the original forced photometry. The data therefore give no measured prediction gain for these fixed curves. They do not establish a clear error difference or exclude a changing repeat pattern or a binary black hole model. The package preserves exact data responses, provenance, source code, numerical checks, and four figures.

AT 2021hdr · ZTF21aaqqwsa · active galactic nuclei · Zwicky Transient Facility · periodicity · prediction · time series · photometry