A constant period fits 127 HAT-P-23 b timing rows from 2008 to 2024 with the stated errors

We tested smooth period change in the hot Jupiter HAT-P-23 b. We measured 56 transit times from public 120-second TESS light curves in sectors 54, 55, and 81. Coverage rules rejected 11 incomplete windows before timing fits. We combined the times with a fixed 71-row historical table. The fitted period change is +0.59 ± 1.36 milliseconds per year under the primary timing errors. The two-sided Gaussian 95% interval is −2.08 to +3.25 milliseconds per year. BIC for the quadratic clock exceeds BIC for the linear clock by 4.66. The 15 events in 2024 agree with the historical linear prediction: their mean difference is −17.7 ± 18.8 seconds, including clock uncertainty. Flux, transit-shape, and trial common-error tests also favor a constant period. The formal period-change error falls by 26% relative to the fixed historical sample, but its lower interval endpoint does not give a stronger decay bound. Removal of the large early source group greatly weakens the constraint. Earlier TESS surveys also found no clear period change. This work adds an independent timing set, a forecast test, and explicit noise controls. The package preserves raw inputs, hashes, selection records, code, covariances, figures, and offline run instructions. The result depends on the historical sample and tested errors. It does not establish orbital decay or a new planet.

HAT-P-23 b · TESS · transit timing · orbital period · orbital decay · reproducible analysis