How we build checklists

Checklist data is only useful if you can trust the numbers. Every figure on this site is traced to the document it came from, and we publish that trail on each set page.

Where the data comes from

We work from official manufacturer checklist workbooks and sell sheets — the same files distributed to hobby shops through authorized distributors. We do not transcribe from other checklist databases, price guides or collector wikis. Each set page lists the source file, its publisher, the date we retrieved it, and a SHA-256 hash of the exact file we parsed.

How it gets processed

  1. Retrieve the official workbook and record its hash.
  2. Parse every row programmatically — no manual re-typing, no sampling.
  3. Reconstruct the hierarchy. Manufacturer files flatten base sets and their parallels into one column. We rebuild the real structure so a parallel is attached to its subset rather than masquerading as a separate set.
  4. Read box configuration and per-box guaranteed hits from the official sell sheet where one was published. Where none exists, the page says so.
  5. Validate before publishing: card numbers must be unique within a subset, print runs must be positive integers, and a parallel whose print run varies card by card is recorded with the full set of values rather than collapsed into a single number.

What we do not publish

Why the CSV comes as two files

Each release exports two files: one row per card, and one row per parallel. They are two different shapes of data. A parallel is a property of a subset rather than of a single card — flattening both into one table would throw away which subset each parallel belongs to, and would repeat every parallel once per card in that subset.

Corrections

Manufacturer checklists get revised after release, and parsing is never perfect. If a number looks wrong, tell us — corrections are published with a dated changelog entry on the affected page.

Currently catalogued: 215 sets, 227,306 cards.