The T16 Planet Hunt: 10,000 New Planet Candidates from TESS Cycle 1 and the Confirmation of a Hot Jupiter Around TIC 183374187

Abstract: The T16 project has produced a uniformly detrended and systematics-corrected set of 83,717,159 TESS Cycle 1 full-frame image light curves for stars observed by TESS in its primary mission down to T=16 mag, enabling sensitive transit searches beyond the official TESS pipelines. While most existing TESS planet searches focus on relatively bright targets, planet occurrence rates suggest that a substantial number of planets should exist around fainter stars. We therefore use the T16 light curves to conduct a semi-automated search for transiting exoplanets across the full Cycle 1 FFI sample, resulting in 11,554 planet candidates orbiting stars down to 16th magnitude in the TESS band with orbital periods between 0.5 and 27 days. Of these, 10,091 are new planet candidates, and 411 are single-transit events, for which we do not attempt to determine orbital parameters. The remaining 1,052 candidates are previously known TESS candidates. We validate our pipeline through Magellan/PFS radial-velocity follow-up measurements on one of our candidate hosts, TIC 183374187, a metal poor thick-disk star, confirming the signal as newly identified hot Jupiter. This detection demonstrates our pipeline's ability to identify real, previously undiscovered, transiting planets. Overall, this work shows that large-scale, machine learning-assisted transit searches of TESS full-frame images can significantly expand the census of transiting planet candidates, particularly around faint stars, providing a rich target set for future validation and follow-up efforts. Our findings more than double the number of known TESS exoplanet candidates.

• The paper presents a statistically robust transit search using TESS Cycle 1 FFI photometry to identify over 11,000 exoplanet candidates, doubling the previous catalog.
• It employs an integrated machine learning pipeline with dual Random Forest classifiers and advanced detrending methods, achieving an F1 score of 0.91 in signal recovery.
• The study confirms a hot Jupiter orbiting TIC 183374187, demonstrating the pipeline’s effectiveness in detecting planets around faint, metal-poor thick disk stars.

Large-Scale Transit Search and Confirmation: The T16 Planet Hunt

Overview and Objectives

The study "The T16 Planet Hunt: 10,000 New Planet Candidates from TESS Cycle 1 and the Confirmation of a Hot Jupiter Around TIC 183374187" (2604.18579) presents a comprehensive, statistically robust transit search leveraging full-frame image (FFI) photometry from TESS Cycle 1, culminating in a catalog of 11,554 candidate transiting exoplanet systems and the dynamical confirmation of a hot Jupiter orbiting a metal-poor thick disk star. The project utilizes 83.7 million detrended light curves down to $T = 16$ mag and implements an integrated pipeline based on machine learning classifiers, contamination checks, and validation through high-precision radial velocities.

Data Reduction and Light Curve Architecture

The core technical advance is the application of large-scale image subtraction photometry to FFIs, pushing transit searches into the regime of fainter host stars. The pipeline employs Difference Imaging Photometry—optimized for crowded fields and blended stellar populations—with photometric apertures anchored to Gaia astrometry, reducing flux contamination and systematics. Detrending is achieved through sequential Spline-based External Parameter Decorrelation (SEPD) and the Trend Filtering Algorithm (TFA), suppressing long-term variability, systematic artifacts, and stellar noise.

Figure 1: Detrending effectively removes large-scale variability, revealing transit signals in the T16 light curves.

Automated Classification and Vetting

To process the immense dataset, event selection is driven by dual Random Forest Classifiers (RFCs) independently optimized for distinct brightness regimes. Light curves are characterized by over 50 diagnostic features focused on S/N, harmonic content, and transit morphology, aggregated from both Cambridge Exoplanet Transit Recovery Algorithm (CETRA) output and refined BLS statistics. The classifiers exhibit high fidelity on test sets: the all-star model achieves an F1 score of 0.91 and recovers 87% of transit signals withheld from training Figure 2.

Figure 2: RFCs robustly distinguish between planetary transits, eclipsing binaries (EBs), and null light curves across brightness regimes.

Contamination is rigorously addressed. Ephemeris and spatial cross-association identifies and flags blending, and difference imaging confirms candidate centroiding Figure 3, eliminating off-center events indicative of background contaminants or astrophysical blends.

Figure 3: Blended sources are detected as off-center residuals in the difference images, helping suppress false positives.

A secondary RFC mimicking manual vetting is applied post-contamination filtering, maximizing recall while further reducing spurious events. Full vetting summary plots—a diagnostic suite examining phase-folded morphology, stellar context, variability, and spatial origin—support the final candidate culling Figure 4.

Figure 4: Vetting plot for a confirmed candidate, compiling photometry, BLS fits, HR diagram context, and centroid analysis.

Catalog Characteristics

The final catalog comprises 11,143 multi-transit and 411 mono-transit candidates, more than doubling the current TESS candidate pool and substantially extending into under-explored faint host and complex Galactic populations. The period distribution peaks sharply at $\sim$3.5 days, consistent with hot Jupiter occurrence, with an SNR-dependent drop-off at longer periods Figure 5.

Figure 5: Candidate period distribution, exhibiting the characteristic hot Jupiter "pile-up" near 3.5 days and a decline for longer periods due to decreasing detection probability and SNR.

Parameter-space analyses show the candidate population spans the Hertzsprung–Russell diagram but is dominated by gas giants, with minor fractions of Neptunians, sub-Neptunes, and a handful of super-Earths (Figures 10, 12). The large majority of hosts are K and G dwarfs, with a subset probing M-dwarfs, F, and even A/B stars.

Figure 6: Distributions and correlations of magnitude, host parameters, and candidate density, with Mollweide projection of Galactic positions.

Figure 7: Hosts distributed throughout the HR diagram; late M-dwarfs and evolved stars present new follow-up opportunities.

Spatially, faint candidates exhibit enhanced detection in crowded regions, demonstrating the effectiveness of the image-subtraction pipeline over aperture photometry techniques Figure 8.

Figure 8: Candidate distribution in Galactic coordinates, with enhanced detections in the bulge for faint sources.

TIC 183374187: Planetary Confirmation

The dynamical confirmation of a hot Jupiter at TIC 183374187 exemplifies the pipeline's yield. The host is a thick disk G-type star with $\mathrm{[Fe/H]}\approx -0.4$ and elevated $\alpha$-element abundances, as confirmed by Gaia DR3 astrometry, ground-based spectroscopy, and kinematic analysis Figure 9.

Figure 9: Toomre diagram: TIC 183374187 kinematics are typical of a thick disk population.

RV follow-up with the Planet Finder Spectrograph on Magellan yields a Keplerian solution with $P=5.06$ d, $K=74$ m\,s$^{-1}$, and $M_p = 0.56\,M_\mathrm{J}$ Figure 10. Simultaneous modeling of transit photometry and RVs constrains $R_p = 1.25\,R_\mathrm{J}$, placing the planet in the hot Jupiter regime with low inferred bulk density.

Figure 10: RV time series and phase curve with best-fit eccentric and circular Keplerian models.

Phase-folded TESS T16 photometry shows robust transit detection, and isochrone placement situates the host as old and slightly evolved (Figures 7, 8).

Figure 11: Phase-folded TESS light curve with best-fit transit model, confirming a deep, periodic signal consistent with a gas giant.

Figure 12: Stellar density and effective temperature position the host at $\sim$12–13 Gyr in the thick disk population.

A full blend analysis rules out hierarchical triple and background EB scenarios at high significance ($\sim$0BIC $\sim$1).

Theoretical and Practical Implications

This work validates the feasibility of large-scale, FFI-based, ML-driven surveys for exoplanet discovery around faint stars, strongly enhancing the statistical sample beyond previous pipeline efforts. The ability to confirm planets in the thick disk opens new parameter space for planet population studies as a function of host metallicity and Galactic history.

The substantial number of faint candidates enables constraints on planet occurrence rates and distributions in under-sampled stellar and Galactic populations, pending follow-up efforts. The utility of difference image photometry—combined with ML models incorporating ephemeris and spatial clustering—effectively mitigates contamination, supporting robust, high-yield candidate identification.

Notably, the confirmation of a hot Jupiter around an old, $\sim$2-enhanced host constrains evolutionary scenarios for gas giant migration and survival in dynamically hot and chemically distinct populations. The catalog's breadth provides targets for probing occurrence variations as a function of metallicity, age, stellar type, and Galactic component.

For future TESS cycles and cross-sector analysis, the scalable framework is poised to recover longer-period, lower-SNR, and multi-transit systems, further extending completeness. Additionally, the large training set of true and false positives will enable future CNN- or transformer-based classifiers to operate on both light curves and direct imaging data, potentially eliminating the need for manual vetting and further improving classification purity.

Conclusion

The T16 project (2604.18579) demonstrates a paradigm for statistically complete, large-scale exoplanet discovery in FFI data, producing a catalog that increases the known TESS candidate count by %%%%13$\mathrm{[Fe/H]}\approx -0.4$14%%%% and confirming a representative hot Jupiter in an old, metal-poor thick disk system. The project introduces methodology for robustly suppressing contamination and controlling false positive rates in crowded, faint source regimes—critical for demographic studies extending to the lowest-mass, oldest, and most distant stellar populations accessible to TESS. The scalability of the pipeline and catalog is directly applicable to the extended TESS mission and future all-sky surveys, enabling systematic characterization of planetary system architecture across the full diversity of Galactic environments.