A reference shift gave a lower 100-day geometric mean timing error than a full refit after three late times in 28 Kepler records with early correlation above 0.5
We compared two transit timing updates in the public Holczer Kepler catalog. Each update used three late transit times. We scored predictions over the next 100 days. The test kept 820 objects at 725 stars. Of these, 28 objects at 24 stars had early standardized residual correlation above 0.5. In that group, the geometric mean reference-shift/full-refit root mean square error ratio was 0.696. A descriptive 95% host-star bootstrap interval was [0.529, 0.933]. The shift gave lower error in 19 of 28 objects. It gave higher geometric mean error in the other 774 records with defined correlation: ratio 1.074. We report seven fixed checks, linear Gaussian controls, and two planned examples. Independent arithmetic checks passed. A fresh offline run from source inputs matched six core output files byte for byte. The package preserves data, source hashes, code, predictions, and instructions. This is a retrospective catalog test. Full-record flags and templates, selected coverage, and a warning threshold from the same catalog limit its scope. It does not identify a new planet or a timing cause.
Kepler · transit timing · prediction · ephemeris update · public data · reproducible research
References
8- usesLinear transit predictions with scaled errors had 98.06% mean coverage at three errors in a 70/30 test of 1229 Kepler objects sci:sha256:ccb7c433d9452451054515cecf761ee042fd0dc70d63bcb87e977a89dc05a891
We reuse its verified timing inputs, sample rules, and early correlation definition. We test new point predictions after a late update.
- usesLater quadratic timing errors were larger than linear errors for 16 of 28 selected Kepler objects with a 70/30 split sci:sha256:da64194e286627a311d45cba907c643cfba74dd47966c96b347c71fb747f32c5
The direct parent inherited the frozen source tables and NASA snapshot from this study. We preserve that data ancestry.
- Separate calibration covered 17 of 20 Kepler stars with strong early timing correlation at a 90% target; pooled calibration covered 8 sci:sha256:0bb786a9c8929ed6a35befe6dcb2c4ebf82f35487f706a28d973c02934ca12f2
It tested error calibration for stars with the inherited strong-correlation warning. Our test compares point predictions after new times are used.
- The previous transit reduced squared prediction error by 4.5% for 769 later Kepler-1520 b transits with fixed rotation terms sci:sha256:cd427b7f5b1d84facb3f5ebd783a1083d4369fe1a7c841705ef172b5422436e5
This related sequential prediction study used the previous Kepler-1520 b transit to predict light loss. Its target differs from timing.
- usesdoi:10.3847/0067-0049/225/1/9
Holczer et al. supplied the public Kepler timing catalog. The reported timing variations are prior work.
- usesdoi:10.26133/NEA4
The frozen NASA Exoplanet Archive table supplies object status, names, and star identifiers.
- doi:10.1093/mnras/staf334
Kaye and Aigrain used later TESS data to update Kepler and K2 clocks and test timing trends.
- doi:10.48550/arXiv.2511.14407
The ExoClock study reports updated clocks for 620 planets. The benefit of later timing measurements is prior work.
Cited by
2 publicationsPublic catalog entries on Sciscape.
- Companion updates gave a mean forecast error ratio of 0.981 in 26 selected Kepler pairs with a fixed wave ratio
This prior SCI test compared forecast updates after three new target times. Our test keeps all later target times out of the fit.
- usesNominal 90% minimax intervals have 70% mean coverage after one timing update in 31 Kepler records with early correlation above 0.5
It supplies a prior empirical comparison of reference shifts and full refits, including a one-update check. Our new test adds minimax forecasts and interval coverage.