Optimized period change intervals are at least 10 percent shorter in 11 of 115 transit systems with 60-second source bounds and fixed Gaussian errors

We tested intervals for quadratic transit timing terms in a fixed 2022 public catalog. The sample has 4,378 times for 115 systems. Each timing source can have a constant unknown shift within a stated bound. We select linear weights to minimize the sum of a Gaussian noise limit and the maximum source bias. The weights do not use the measured timing residuals. At a 60-second bound per source, 11 of 115 systems have intervals at least 10 percent shorter than the better of two endpoint estimates. These endpoints use a common-intercept estimate with the same bias allowance, or free source intercepts. The median reduction is 0.24 percent. The largest reduction is 18.84 percent. Tests with other source labels and errors also give small median gains. The interval coverage proof assumes independent Gaussian errors with fixed, known standard deviations. The primary intervals use a family correction for 115 systems and seven bounds. Even with free source intercepts, 38 systems have reduced chi-square above 2. Thus a nonzero interval does not establish orbital decay. We supply 3,178 intervals across four analysis choices, all estimator weights, source bias contributions, input provenance, figures, code, and reproduction checks. The contribution is a tested method and sensitivity dataset, not a new planet or a new decay detection.

transit timing · orbital period change · source offsets · bounded bias · Gaussian errors · confidence intervals · convex optimization · reproducible astronomy