Interpreting Results

Detection, ranking, and validation are different

Layer	Question answered	Does not mean
BLS peak	Is there a strong periodic box-like signal in this grid?	The signal is planetary
RF/CNN score	Does this peak resemble prioritized training examples?	Calibrated planet probability
Preliminary pass	Did simple shortlist checks avoid a configured failure?	Independent validation
Weak candidate	Did the signal avoid key failures but lack complete strong evidence?	Confirmed planet
Strong candidate	Did all configured evidence requirements pass?	Physical confirmation

Read empirical FAP correctly

FAP = 0.01998 means that under the chosen segment-shuffle null and complete BLS grid, 19 of 1,000 null target maxima met or exceeded the observed power; the plus-one estimate is 20/1,001. It is conditional on this test and does not say that the object has a 1.998% probability of being a false positive.

Understand catalog absence

The search excludes known catalog history before signal analysis. A later non-match only means no matching row was found in the queried tables at that time. Catalogs have coverage, naming, timing, and sensitivity limits. Absence does not establish novelty.

Understand external evidence

• a clean Gaia scene reduces one contamination concern but does not eliminate unresolved blends;
• no TESS match may reflect non-observation, cadence, dilution, or sensitivity;
• an ExoFOP non-match is not a validation result;
• a U-shaped fit is compatible with a transit but not unique to planets; and
• a plausible radius depends on stellar parameters and model assumptions.

Report SXS outputs responsibly

Use precise labels such as:

SXS identified an unconfirmed weak transit-like signal requiring independent follow-up.

Do not write:

SXS discovered a planet.

Every shared table or figure should retain target identifier, candidate label, config/release version, and the unconfirmed-candidate warning. Publications must also follow the citation and license requirements.