Landing Page
Landing Page
Card Image Upload Screen
Initial Card Scan to Identify Card Screen
Result Screen

GradeRef: Reinventing Sports Card Grading

Inspiration

Sports card collecting is a 12.9 billion dollar industry, with a continuely growing market, appealing to a wide array of people. This should be a fun and rewarding hobby, but the current grading system is broken. Third party companies like PSA are in charge of grading the quality of these cards, but with inconsistent and non-transparent methods, they can affect the valuation by as much as 8x. With factors such as:

Grading is expensive ($25 per card minimum).
Regrading is risky and time-consuming (> 45-day turnaround).
Low-value cards are excluded from the market because grading isn’t financially viable.

However, PSA remains one of the few widely trusted third-party services for grading card quality. From this, we saw an opportunity: What if grading was accessible to everyone? Instead of blindly trusting an arbitrary system, collectors should have the ability to scan their own cards and receive instant, data-driven predictions—eliminating the guesswork and leveling the playing field.

What It Does

Our AI-powered tool allows collectors to scan a sports card and receive a predicted PSA grade instantly.

Uses machine learning trained on a dataset of thousands of PSA-graded cards.
Recognizes PSA’s grading patterns, despite its subjectivity.
Helps collectors determine whether a card is worth sending for grading.
Eliminates the risk of paying high grading fees without certainty.

By providing instant grading estimates, our solution ensures that all collectors, not just those who can constantly afford PSA’s steep fees, can participate in the hobby.

How We Built It

Collected a dataset of thousands of PSA-graded sports cards, labeled with their official grades.
Preprocessed images to account for lighting, angles, and wear patterns.
Developed a convolutional neural network (CNN) trained to recognize grading characteristics.

Challenges We Ran Into

Data availability: Finding high-quality PSA-graded card images with consistent metadata was a challenge.
Subjectivity of PSA grading: Since PSA’s system isn’t standardized, we had to reverse-engineer their grading patterns using data-driven methods.
Cropping accuracy: Ensuring that the app could find the card given different lighting conditions and camera quality variations.
Model training time: Processing and refining the dataset took significant time and computational power.

Accomplishments That We're Proud Of

Successfully trained a machine learning model that can predict PSA grades with high accuracy.
Developed a system that predicts grading, making sports card collecting more accessible.
Created a tool that saves collectors money and time, reducing uncertainty in regrade requests.

What We Learned

Creating and cleaning datasets is a lot harder than we thought, especially when it came to images.
AI can bridge the gap between human subjectivity and consistent standards.
The market is hungry for alternatives to expensive, slow, and inconsistent grading services.

What's Next for GradeRef

Expanding the dataset to improve accuracy across multiple card conditions and brands.
Enhancing mobile scanning capabilities for better real-world usability.
Integrating support for autographs, memorabilia, and other collectibles that rely on subjective grading.
One day, replacing PSA as the industry standard—creating a world where collectors no longer need to rely on an outdated, unfair system.

Grading should be accessible. The hobby should be for everyone.