Early sinus fits give no stable later gain for three proposed KIC 10001893 planet signals in Kepler release 25

We test three known low-frequency signals of KIC 10001893 with public Kepler release 25 light curves. We fit frequencies and sinus coefficients in Q3 and Q5 to Q10, then predict Q11 to Q17. Across two flux products and three trend scales, 17 of 18 fits increase later squared error. All quarter-resampling intervals include zero. The central 14.26 microhertz fits keep a similar phase but give early amplitudes of 46 to 48 ppm and later amplitudes of 20 to 21 ppm. A correction with archive crowding factors does not restore positive gain. In real-flux injection tests, positive gain occurs in 16 of 24 trials at 20 ppm and all 24 trials at 40 ppm. These are conditional sensitivity tests. They are not planet exclusion limits. The results do not confirm a planet, prove signal change, or identify a physical cause. Both products share exposures. Independent instrument confirmation is absent. The bundle includes original code, aggregate results, a figure, exact source hashes, and external retrieval steps.

astronomy · KIC 10001893 · Kepler · planet candidates · chronological prediction · weak signals · null result