A 1-to-30-day Kepler-90 box search misses known planet i and gives no later pass for 20 residual peaks

We search Kepler release 25 light curves of Kepler-90 for extra transits with periods from 1 to 30 days. We mask all eight known planets. We select 20 box-search peaks in quarters 1 to 9 and test their fixed periods, epochs, and durations in available quarters 10 to 17. 0 of 20 residual signals pass a later shifted-phase rank threshold of 0.0025 with at least three sampled events. The test uses 999 shifted phases for each signal. Known planet i gives depths of 83 and 98 parts per million at its published ephemeris in the two time intervals. A blind search with i left unmasked does not select its period among the 20 peaks. This is a sensitivity failure. Conditional injections test later fixed-ephemeris sensitivity. They do not measure blind-search completeness. Missing quarters, masking, period error, correlated noise, and detrending limit the result. The study does not set a planet occurrence limit. No independent observatory confirmation is available. Code, fitted test results, figures, source hashes, and retrieval steps are provided.

astronomy · Kepler-90 · transit search · time validation · null result · box least squares