Stursberg, K. (2026). CryptoTaxEdge [Dataset]

An open, versioned classification schema for blockchain transactions, together with the labelled benchmark corpus it was measured on. The schema (CryptoTaxEdge Classification Standard v1.1.0) defines a category vocabulary, a closed five-value tax-treatment enum with US framing, a doctrine for categories that have more than one defensible tax position, confidence and review semantics, evidence standards, and a JSON wire contract that a classification record can be validated against. It is published with worked examples and with the tax-position caveats attached: a category and its default treatment are a starting point for professional review, not a tax position, and several categories carry documented alternatives a firm can elect instead. The corpus is the 388-transaction adjudicated answer key from the 2026 H1 bake-off study: public transaction hashes on Ethereum, Polygon, and Arbitrum, each with a category label and the schema default treatment for that category. Labels were built under a protocol pre-registered before execution, from two independent blind model lenses reading raw receipts and logs. The 96-transaction complexity-ladder hash list is included; that study has no ground-truth labels by design.