Datasheet
What this dataset contains, how it was made, how complete it is, and what it will not do — so you can judge it without reading it.
What it is
| Subject | Credit denominated in dollars against bitcoin collateral, across every venue class we can observe. |
|---|---|
| Unit of observation | A venue. Each is described by the same criteria whatever its class. |
| Classes covered | On-chain protocols (pooled, isolated, CDP), auction venues, CeFi desks, the corporate layer, brokers, derivatives. |
| Criteria | Three pillars — Price, Quality, Composition — published and versioned before any venue was measured against them. Methodology. |
| Per-cell provenance | Every criterion value carries the kind of source it came from, the source, and an as-of date. |
| Last read | 2026-09-05 |
How complete it is
| Venues in the registry | 168 |
|---|---|
| Publishing anything yet | 12 |
| Criteria fully researched | 16 of 168 venues are 8 of 8. The rest carry unknown (reason) in the cells not yet established. |
| Contributing to the surface | 15 of 21 (71.4%) |
| Named as absent | broker_fidelity, broker_ibkr, broker_schwab, cefi_ledn, cefi_xapo, maple_ethereum |
Limitations, stated plainly
- Coverage is partial and uneven. Only 16 of 168 venues have every criterion researched. A venue with unresearched cells is not a venue with nothing to report — it is a venue we have not finished reading.
- Some series are narrower than the market. The composition and flow series are computed from Morpho Blue only, on Ethereum and Base, and say so in their own payloads. They are not market-wide and must not be quoted as such.
- Absence is not zero. A venue missing from a reading is a declared gap in observation. We name the absentees rather than letting a percentage imply they were measured and found empty.
- Classes are not like-for-like. A fixed-term auction rate and a variable pooled rate are different objects. Cross-venue rows carry a caveat saying so; read it before placing them in one column.
- Source kinds differ systematically by class. An on-chain venue's admin map is checkable; a CeFi desk's terms are its own statements. That difference is recorded per cell, and is itself one of the deepest comparisons here.
What it refuses to do
No composite score. No ranking. No default ordering by rate. No safety verdict on any venue. None of these will be added — they are not missing features. A reader weighs the criteria; this dataset does not weigh them for the reader, and a single number would make it a benchmark rather than a description.
How to get it
| MCP (read-only) | https://mcp.bitcoincreditstack.com/mcp — 12 tools, zero write tools. |
|---|---|
| REST | https://api.bitcoincreditstack.com/ |
| Endpoint index | /endpoints.json |
| Machine index | /llms-full.txt — every venue, by class. |
| Sitemap | /sitemap.xml |
Re-use
The licence is drafted and with counsel — see Legal. Until it is published, no licence is asserted and none is implied. There is deliberately no machine-readable license field in this page's structured data either: asserting one in schema before publishing it would be publishing it.
Canonical: https://www.bitcoincreditstack.com/data