TESS flux from 2021–2024 gives a TOI-2337 b period change of −0.29 ± 1.32 seconds per year with fixed noise estimates

We test a steady change in the orbital period of TOI-2337 b. We use 166 complete transit windows from 28 public TESS SPOC light curves with a 120 s cadence, from 2021 June to 2024 December. We fit a common transit shape and orbital clock directly to the flux. Each window has a local polynomial baseline. Noise estimates from samples outside transit remain fixed. The fitted period derivative is −0.29 ± 1.32 seconds per Julian year, where the error is one local standard error. A quadratic clock reduces chi-square by only 0.048 relative to a linear clock. Holding the rate at the published +9.828 seconds per year increases chi-square by 57.1 after the other parameters are fitted again. This test applies to a constant rate across the full interval. The earlier-span flux test alone gives +6.48 ± 7.68 seconds per year and cannot distinguish zero from the published rate. A sector timing check agrees with the direct flux result. Gaussian summaries of individual event times give excess scatter and a larger apparent quadratic gain. The full event likelihoods show why that summary can fail. A separate regression on all 102 published TOI-2337 times does not reproduce the reported rate; the original local timing cuts are not reconstructed. We preserve input arrays, archive hashes, selection records, source code, results, and reproduction instructions. The result depends on a constant symmetric transit shape and the stated noise assumptions. It is not a detection of physical orbital decay.

TOI-2337 b · TESS · transit timing · orbital period · evolved stars · correlated noise · reproducibility