Free, machine-readable structured-finance data — derived entirely from public SEC EDGAR filings, with provenance on every value.
Dealcharts publishes open datasets for anyone building on structured-finance data: per-deal CSV exports, per-entity facts JSON, an agent manifest (
), and externally-hosted machine-learning benchmarks. Every figure traces back to a primary SEC filing and can be reconstructed from source. Licensed CC-BY 4.0 unless a specific dataset says otherwise; AI training and retrieval are explicitly permitted.llms.txt
Current public export surface, as of July 2026: about 1,230 CMBS and Auto ABS deal CSVs, 41 BDC CSVs, three ML dataset descriptors, a provenance sidecar at
, and file discovery through /datasets/_meta.json
./sitemap-datasets.xml
Who it's for: quantitative researchers, data scientists and ML engineers, credit and CRE analysts, and AI agents that need citable, disclosure-derived facts rather than scraped guesses.
Jump to the data:
Need the read for a specific position? Email cairn+llms@cmdrvl.com — name the CUSIP or seat and Cairn writes back with the position's read and the source receipts.
Leakage-safe, SEC-derived benchmarks for machine learning on CMBS loan performance. Live research copies are free on Hugging Face under a non-commercial license; commercial use is licensed directly through CMD+RVL. Descriptor-only shapes are published with ODPS metadata before the data files go live. For access or a commercial license, email cairn+llms@cmdrvl.com.
Leakage-safe flat benchmark: predict which CMBS loans transfer to special servicing within 12 months, using only point-in-time SEC-disclosed features. For TabPFN, AutoML, XGBoost, and credit-risk work.
Per-loan event-sequence shape for transfer timing models and temporal credit-risk research. Data file in preparation; descriptor and ODPS metadata are public.
LLM-native DuckDB research pack for local analytics, citation-ready retrieval, and transfer-risk experiments. Data file in preparation; descriptor and ODPS metadata are public.
Dealcharts organizes public filings, deals, and counterparties into datasets you can explore, cite, and query through public facts endpoints.
Each dataset carries provenance — source, author, timestamp — ensuring every point can be traced to where it came from.
| Dataset | Description | Source | Update Frequency | Status |
|---|---|---|---|---|
| CMBS Deals | Details of commercial mortgage-backed securities deals, shelves, tranches, and counterparties. | SEC EDGAR (424B, 10-D) | Weekly | Live |
| ABS Auto Deals | Auto loan and lease securitizations, with performance and delinquency metrics. | SEC EDGAR | Weekly | Live |
| Fund Holdings with CMBS Exposure | Crosswalk of mutual fund holdings mapped to CMBS tranches. | NPORT-P Filings | Monthly | Live |
| Entity Registry | Issuers, servicers, trustees, funds, and sponsors normalized across filings. | Derived | Rolling | In Progress |
| Verification Events | Records of confirmed profiles, contributor links, and corrections. | Dealcharts internal | Continuous | Pilot |
| Context Creation Corpus | Verified explanations and rationales captured with provenance. | Verified submissions | TBD | Planned |
Verification fees support continued open access and coverage expansion.
Data originates from public sources (e.g., EDGAR) and curated references.
Static CSV exports are generated at build time for deal and BDC records.
Facts JSON carries the machine-readable per-entity shape for deals, BDCs, funds, and CLOs.
documents source, license, URL patterns, counts, and CSV columns./datasets/_meta.json
and/datasets/{deal-key}.csv
./datasets/bdc-{ticker}.csv
,/llm/facts/{deal-key}.json
, and/llm/facts/bdc-{ticker}.json
./llm/facts/fund/{fund-key}.json
Download individual deal files in the formats that exist today:
- generated per-deal CSV. CMBS files include pool metrics; Auto ABS files include deal identity fields until ABS metric exports are available./datasets/{deal-key}.csv
- generated per-BDC CSV from public filings./datasets/bdc-{ticker}.csv
- machine-readable facts for LLMs, APIs, and citation workflows./llm/facts/{deal-key}.json
- source, license, counts, URL patterns, and CSV column dictionaries./datasets/_meta.json
- schema-ready JSON contract for newly watchlisted and removed-from-watchlist loan rows. Current row count is zero until upstream watchlist flags are exported./datasets/cmbs-watchlist-transitions.json
https://dealcharts.org/datasets/wfcm2025-c65.csv
https://dealcharts.org/llm/facts/wfcm2025-c65.json
https://dealcharts.org/datasets/_meta.json
Attribution required: "Source: Dealcharts.org" with link to page.
Datasets may not include personal or confidential information.
Learn more in Disclosures and Privacy.
Public paths that work now are documented at
. For a custom bulk pull or a commercial package, email cairn+llms@cmdrvl.com with the CUSIPs, funds, issuer, asset class, or date range you need./api/