A circular test in five period windows gives a 3.9% Gaussian noise tail probability for Boyajian's star velocities

We test the public SOPHIE radial velocities for Boyajian's star, KIC 8462852. We use 43 measurements from 29 nights in Table B1 of Madurga-Favieres and colleagues (2026). The test assumes a circular orbit, fixes the reported transit time, and searches the five initial period windows in that paper. It fits a velocity offset and extra independent Gaussian noise between nights. The best signal gives an increase of 5.093 in twice the log likelihood. In 10,000 simulations from the fitted null model, 387 results are at least this large. The conditional tail estimate is 3.88%, with a 95% interval of 3.50% to 4.27% from simulation error alone. This gives modest support under the stated assumptions. The maximum fitted likelihood ratio between the five windows is 3.73. Removing one night changes the test statistic from 3.391 to 7.238. Signals with the source paper's amplitude of 162 m/s are detected in 26% to 61% of trials, depending on the window, at the conditional 5% threshold. These new calibration and power results show limits set by the present sampling and noise. They do not confirm a planet or reproduce the source paper's eccentric joint fit. The noise level is estimated, so the calibration is approximate. The simulations do not include the source study's selection of the period windows or its full search history. The package preserves the source article, data extraction, code, numerical checks, simulation results, and four figures.

Boyajian's star · KIC 8462852 · SOPHIE · radial velocity · planet candidates · parametric bootstrap · detection power · circular orbit