Calcium interpolated 4.40 days later lowers error in an exploratory HD 26965 NEID velocity test with free season offsets

We test whether a calcium time shift transfers between HD 26965 observing seasons. Public NEID survey tables give 157 spectra and 83 night averages. A linear fit uses the first season alone. Each later season has a free velocity offset. The planned rule uses calcium at t - 6.66 days. On 40 later nights, it raises root mean squared error from 1.3934 to 1.5670 m/s. An added audit of the 2024 published table compares both shift directions. On 56 common spectra, line-by-line velocity correlation is 0.704 for calcium at t + 6.66, 0.313 at t, and -0.366 at t - 6.66. The pipeline series instead favors simultaneous calcium among these rules. The added catalog test uses one common sample for all shifts from -10 to +10 days. Training error selects calcium at t + 4.40 days. On 36 later nights, this rule reduces error from 1.1766 to 1.0902 m/s. Mean squared error falls by 14.15%, with a lower measured error in both test seasons. This added analysis is exploratory because the planned result was already known. Conditional 95% block-bootstrap intervals include zero gain for every tested block length. Later calcium and fitted offsets are required. This is a test of a linear rule, not a physical delay measurement or a planet claim. The package preserves source tables, plans, independent checks, figures, and offline reproduction code.

HD 26965 · NEID · stellar activity · radial velocity · calcium · time shift · season transfer · reproducible research