Methodology

How Grade Potential is estimated

We think you deserve to know exactly how the numbers are produced. Here is the honest version — including the limits.

Deterministic geometry (OpenCV)

Centering and geometry are measured deterministically from the captured image — border ratios, framing, and alignment — using classical computer-vision techniques, not guesswork.

Classical CV heuristics

Corners, edges, surface, and whitening are assessed with established image-processing heuristics. These flag likely condition issues the camera can see.

Versioned conservative rules engine

A versioned rules engine combines those signals into an estimate. It is deliberately conservative and every result records the rules version used.

Ranges with confidence

The output is a range with a Confidence Level — not a single number and not a promise. Lower-quality captures widen the range and lower confidence.

A general language model is never the sole engine

The analysis is grounded in deterministic geometry and classical computer vision plus a versioned rules engine. A general-purpose language model is never used as the sole grading engine, and it never overrides the measured signals. This keeps results explainable and reproducible for a given rules version.

What the camera can and cannot see

A camera captures what light and focus allow. It can miss microscopic surface scratches, hidden print defects, and issues obscured by glare, sleeves, or soft focus. Professional grading is subjective and performed in person, so a grading company may reach a different conclusion. That is why every estimate is a range, and every report carries a disclaimer.

The language we use

To keep things honest, we only ever describe results with these terms:

Grade PotentialEstimated PSA RangeEstimated Grade CeilingComputer-Vision Condition AnalysisPossible PSA 10 CandidateSubmission CandidateEstimated Maximum GradeConfidence Level

We never make grade guarantees or claim an official or approved grade. Grading-company names are used only to describe the scale a range refers to.

The disclaimer on every report

This is a computer-vision estimate, not a grade. Professional grading is subjective and performed in person; cameras can miss microscopic or hidden defects, and lighting, focus, glare, sleeves, and image quality affect results. Pokémon Stock Radar is not affiliated with PSA or any grading company, and a grading company may return a different result. This is not financial advice.

See the Grade Potential workflow Data attribution