Early 32.761-day waves reduce later GJ 536 fit error by 16 to 21 percent under a fixed rotation model

We test time transfer at the published 32.761-day period of GJ 536 c in HARPS RVBank version 02. Waves fitted to 130 early exposures reduce conditional weighted fit error in 65 later exposures by 15.85 percent for SERVAL and 21.36 percent for DRS under a fixed rotation model. Nominal 30-day block ranges are -7.06 to 31.25 percent and -2.72 to 35.02 percent. They include zero. Point gains stay positive with other nuisance terms, the M2 mask, and period shifts up to three quoted errors. The known inner-planet control gives larger gains. The source periods use overlapping observations, nuisance terms are fitted in the later data, and both reductions share spectra. This is a conditional coefficient-stability check, not independent planet confirmation or a blind discovery. No new planet or physical effect is found. Code, aggregate results, a figure, exact input hashes, and external-data retrieval steps are included.

astronomy · GJ 536 · HARPS · radial velocity · time prediction · coherence · activity · uncertainty