Other TESS noise windows give wider timing errors for TOI-6667.01 under a trapezoid model in sectors 1, 28 and 104

We test a reported timing mismatch for TOI-6667.01 with public TESS light curves. We fit three transits at native cadence with separate symmetric shapes and local baselines. The timing contrast is 14.81 minutes in corrected aperture flux and 14.09 minutes in aperture flux. Formal independent-noise errors give 3.60 and 3.37 standard errors. We add model transits to 70, 67 and 70 other windows in each flux series. We match these windows to the observed sample schedules. The contrast error standard deviations are 4.69 and 5.04 minutes. The formal errors are 4.12 and 4.18 minutes. Of 328,300 combinations, 507 (0.154%) in corrected flux and 1377 (0.419%) in aperture flux have an absolute timing error at least as large as the measured contrast. The windows overlap, so these fractions do not give calibrated false-alarm probabilities. We also run 1000 local residual-block trials for each flux series and each of three block lengths. Their absolute-error tail fractions range from 0.0% to 0.2%. These tests assume stationary fitted residuals. The mismatch remains of interest, but its significance depends on the noise treatment. The target was selected after earlier timing reports. This study does not establish a physical timing change or confirm a planet. The package contains the source FITS files, hashes, code, all trials, figures, and offline repeat steps.

TOI-6667 · TIC 161169240 · TESS · transit timing · correlated noise · signal injection · reproducible photometry