Dated fits reduce median TESS transit-time offsets from 6.63 to 0.74 minutes for 129 hot Jupiters

We tested whether a source choice can explain large offsets in published TESS transit times. We reused a fixed sample of 153 hot Jupiters from an earlier SCI study. We selected a period and epoch pair from one NASA Exoplanet Archive row. Its paper had to predate the tested sector. We ranked rows by a prediction error bound, without using the residual. The bound allows every period-epoch covariance consistent with the stated marginal errors. A complete dated fit exists for 136 planets. Their median absolute offset falls from 6.24 to 0.67 minutes. The planned selection with explicit BJD-TDB epoch labels has 129 planets. Its median falls from 6.63 to 0.74 minutes, and flags beyond three maximum error scales fall from 25 to 11. These are conditional error-scale scores, not calibrated detection probabilities. Independent smooth-model fits to five saved raw light curves change their mean times by at most 7.1 seconds. A source check shows that the large WASP-76 b residual depends on which fit in one paper is used. HAT-P-7 b has a previously reported timing trend. This is a retrospective reanalysis of inherited times. It establishes no new planet or period change. We preserve source records, the full archive response, raw diagnostic data, code, figures, and an independent numerical checker.

TESS · hot Jupiters · transit timing · ephemeris · covariance · NASA Exoplanet Archive · reproducible research