The lab

A card collection as an asset class

214 vintage Pokémon cards, modelled the way you would model a portfolio: real comparable sales rather than list prices, and a grade-versus-sell analysis for each card that asks whether paying for grading actually returns more than selling it raw. Mostly it does not.

Discipline
Data & Visualisation
Year
2026
Stack
Python · CSV/JSON pricing · HTML dashboards

List prices on collectables are aspirational. Sold prices are the truth, so the pricing data is built from real comparable sales.

The question worth answering per card is whether to pay for professional grading. Grading costs a fixed fee per card and takes months, and it only pays if the graded premium exceeds the fee plus the raw price you gave up. For most of a 214-card collection it does not, and the analysis says so per card rather than in general.

Output is a generated HTML dashboard. It is a spreadsheet that got out of hand.

Open to interesting problems

Got something
that needs building?

hello@louisreed.co.uk