TOI-270 d

What the corrected fit shows

The timing-adjusted transit is strongly preferred by ΔBIC = 301.7. Its fitted midpoint is +0.343 hours from the historical prediction; the model's mid-transit depth is 2676.0 ± 181.6 ppm. The support decision uses ΔBIC ≥ 10; timing-search p-values are not presented as blind-detection probabilities.

Explicit model comparison

The null fits a local intercept and slope. The transit adds bounded midpoint, radius-ratio, and impact-parameter freedom while retaining the same baseline terms. Fixed limb darkening and circular geometry make this a diagnostic fit rather than a full system retrieval.

Multi-sector robustness and noise

The archive prediction was timing-adjusted independently in 3 fitted sector(s) (S3, S4, S5), of which 3 meet Delta BIC >= 10. Formal depth errors were inflated by sqrt(max(reduced chi-square, 1)) times the residual time-averaging beta factor (observed range 1.00-3.20). The robust inverse-variance model depth across supported sectors is 3164.7 +/- 135.2 ppm; Cochran Q = 0.85 for 2 dof (p = 0.6544). These scaled errors address underestimated scatter and short-timescale correlation, but they are not a full Gaussian-process or physical limb-darkened transit fit.

Model-independent period and depth cross-check

Astropy's BoxLeastSquares checks the physical fit with a box-shaped signal across all three committed sectors. Cadences around the saved archive transit windows of TOI-270 b and c are masked first; otherwise twice planet c's 5.66-day period creates a strong alias near 11.32 days. The disclosed ±5% search around planet d is a consistency test, not a blind discovery search.

The companion-masked box depth agrees within 1.1σ with the independent-sector limb-darkened combination (3164.7 ± 135.2 ppm). The BLS period is only 0.0325% below the archive value and is not used to revise the authoritative ephemeris.

Machine-readable values: bls_crosscheck.csv. Saved mask inputs: companion_ephemerides.csv.

The system

Values are from the exact saved NASA Exoplanet Archive pscomppars row in data/system_parameters.csv.

Data and reproducibility

The observed photometry is the public MAST file tess2018263035959-s0003-0000000259377017-0123-s_lc.fits, TESS Sector 3, collection DOI 10.17909/t9-nmc8-f686. It is stored unmodified. Exact archive URI, URL, retrieval date, and checksum are in data/SOURCE.md.

Run python scripts/analyze_transit.py, python scripts/analyze_multisector.py, and python scripts/analyze_bls_crosscheck.py to regenerate the figures and CSVs, then pytest tests/ -v to check the documented values against the real file.

Limitations

• The orbit is circular and the quadratic limb-darkening coefficients are fixed representative values.
• This is a lower-margin supported result within this portfolio. Its support is more sensitive than the very large-ΔBIC cases to the fixed outlier rule, fitting window, and limb-darkening coefficients.
• The scaled semi-major axis comes from saved composite values whose uncertainties are not propagated.
• The midpoint is searched within a bounded window; ΔBIC is used as the support gate.
• PDCSAP processing, dilution, stellar variability, timing variations, and covariance can bias the geometry.
• Published global fits with physical priors, instrument-specific detrending, and simultaneous sectors remain authoritative.

References

1. Günther et al. 2019 — discovery reference listed by the NASA Exoplanet Archive.
2. Ricker et al. (2015), Transiting Exoplanet Survey Satellite (TESS), doi:10.1117/1.JATIS.1.1.014003.
3. TESS Team, TESS Light Curves — All Sectors, MAST, doi:10.17909/t9-nmc8-f686; Sector 3 used here.
4. NASA Exoplanet Archive, saved TAP row retrieved 2026-08-15.