Run offsets leave 3% of signal information averaged over phase at 172.5 days in short Teegarden data with fixed noise weights

We measure the information lost to free velocity offsets in public observations of Teegarden's Star. The test uses the 467-row CDS radial velocity table from Dreizler et al. (2024), with 370 inverse-variance daily bins and fixed reference noise weights. At 172.5 days, separate offsets in the short ESPRESSO and MAROON-X runs leave 2.9608% of the signal information averaged over a uniform phase. The reference has exactly known offsets. This ratio uses observation times and errors, without measured velocities. It remains near 3% under the tested changes in noise weights and channel correlation. For the phase fitted to CARMENES, HPF supplies 83.4% of the remaining external information after offsets alone. A fixed-period circular transfer fit gives a gain of -0.052 with a conditional standard error of 0.381 on the CARMENES 172.5-day curve. The planet b and c controls give gains close to one. With a known training curve and the reference Gaussian noise model, the external test has only 42.37% power at a two-sided 3-sigma threshold if gain is one. Earlier work already reported non-detection without CARMENES and the short-run offset problem. This contribution gives numerical information loss, power, and reproducible checks. Period, trend, and noise choices change the apparent mismatch. The audit does not establish a planet or a physical cause. The package preserves code, derived results, figures, source provenance, and a hash-checked source download program.

Teegarden's Star · radial velocity · exoplanets · instrument offsets · Fisher information · public data · signal transfer · detection power · reproducible research