Four fixed-period HD 215152 curves lower mean error across 12 omitted seasons with equal season weights

We test whether each reported HD 215152 circular velocity curve improves prediction for complete seasons omitted from fitting. We use all 373 public HARPS rows and fixed periods of 5.75999, 7.28243, 10.86499, and 25.1967 days. Each of 12 seasons is omitted in turn. The primary model selects a noise shape from training rows and uses a smoothed activity indicator. With equal weight per season, adding the four curves gives mean squared error gains of 0.420, 0.236, 0.154, and 0.427 (m/s)^2. Each gain is also positive in five planned controls. The result depends on the score and assumptions. With equal weight per row, the 7.28-day curve has a small gain of 0.030 (m/s)^2. An added check that refits periods within source-derived bounds on training rows changes this gain to -0.017 (m/s)^2. Some individual seasons have negative gains. The new contribution is a reproducible set of season scores, predictions, covariance matrices, and checks for this public record. Earlier work already reports the planets and tests signal stability. Source periods and processing use the full record, so this is a retrospective audit, not independent planet confirmation. The package supplies source addresses and exact hashes, a verified downloader, code, results, a figure, and reproduction instructions.

HD 215152 · radial velocity · exoplanets · season prediction · model validation · HARPS · period sensitivity