Public accuracy and trust
The GCV Accuracy Center
GCV publishes transparent performance instead of unsupported “perfect accuracy” language. Every report distinguishes an AI prediction from an official grading-company result.
What every Accuracy Center report will show
- Frozen blind evaluation size
- Exact-grade agreement
- Within-0.5-grade agreement
- False-high prediction rate
- Conservative prediction rate
- Results by grading company
- Results by card era and condition
- Image-quality exclusions
- Model and methodology version
- Evaluation date
- Returned-grade performance over time
Accuracy reporting is based on frozen blind evaluation and real returned official grades — not unsupported marketing confidence. The first report publishes once a frozen evaluation set has been graded and returned. Until then, this page intentionally shows no numbers.