Distinct debt issuers
11,945
Union of the 8-K and fund-held tables
MARS · Coverage Brief · Corporate Debt
Four substrates, one entity graph: private BDC lending, issuer-disclosed obligations, fund-held bonds, and balance-sheet leverage — each sourced straight from SEC filings, cross-linked to one company graph, and honestly benchmarked against FactSet.
Debt coverage isn't one table — it's four purpose-built substrates, each answering a different question, all resolved onto the same canonical company graph so they compose.
~14,400 dated loan-origination events across ~6,300 private middle-market borrowers, parsed directly from 21 public BDCs' quarterly Schedule-of- Investments filings — the loans a bank or a public bond market would never see, because the borrower is private. Includes seniority, spread, benchmark rate, maturity, and a per-quarter mark (fair value ÷ cost) that catches real distress before it's news: a cross-lender view lets us see the same private credit sitting in multiple BDCs' books on staggered reporting dates.
53,033 debt obligations across 6,537 companies, sourced from SEC 8-K Item 2.03 — a company's own real-time disclosure that it just created or amended a credit facility, term loan, or note. Instrument type, lender, rate, maturity, and secured/ seniority status, extracted the same day it's filed. This is the earliest possible signal on a new debt event — often weeks ahead of the 10-Q where the same facility would otherwise first surface.
8,401 issuers, ~400K issuer-quarters, a maturity wall of ~1.18M rows — built from what SEC-registered mutual funds report holding (N-PORT). Per issuer: bond count, fund-held value and par, weighted coupon, refinancing pressure by year. Layered on top: 14.4 million bond-month price points from independent fund marks, 6.3 million of them genuine multi-fund consensus with a real cross-fund disagreement signal — most informative on exactly the illiquid names where a single evaluated price is itself a model guess.
Aggregate total debt, one row per company-quarter, extracted from XBRL financial statements alongside cash, equity, and the full leverage picture. Not instrument-level, but universal — every public filer, every quarter, point-in-time correct (flagged if a figure was later restated), so it's the right layer for a trend line or a leverage screen even where the debt-stack detail above hasn't been built out yet.
Against a competitive benchmark FactSet itself supplied for this comparison, our per-issuer counts individually match or exceed their own cited figures on two of the rows that matter most:
| Metric | FactSet cites | We measure | Verdict |
|---|---|---|---|
| Fund-held corporate bond issuers | 8,400 | 8,401 | at parity |
| 8-K debt-obligation events | ~22,000 | 53,033 | 2.4× |
| Fund-mark bond-months | 14.4M cited | 14.4M (6.3M consensus) | matches, with nuance |
Historical depth on the 8-K stream runs 2016–01–04 through 2026–09–03, following a full backfill of the 2016–2022 window; pre-2016 coverage is scoped and on hold pending an upstream index.
Two capabilities where a sophisticated buyer should expect a straight answer, not a sales gloss:
| Capability | FactSet | Us |
|---|---|---|
| Debt capital structure | 75,000+ entities linked to 45,000+ ultimate parents. | Strong raw issuer counts (above), but only ~23% of our own parent-linkage gap is closed so far — FactSet ahead on entity-graph depth, not on raw coverage. |
| Bond pricing & terms | Global, daily evaluated prices via a third-party pricing partnership (S&P). | Monthly cadence, US-fund-held bonds only, no independent evaluated-price methodology — FactSet ahead, and our own data says so. |
Of an original $462B in fund-held debt value with no resolved ultimate parent, two independently-built passes — a filing-co-appearance model and a subsidiary-name extraction from other companies' own SEC exhibits, both verified against live web search rather than taken on model confidence alone — have resolved $88.0B (23.3%) of the remaining $378B gap. The two methods independently converged on identical parents for six overlapping issuers, which is real evidence the approach generalizes rather than a lucky coincidence on one pass.
Filing co-appearance — same-document relationships (parent, sibling SPV, underwriter, merger counterparty), verified: 163 resolved parent links, $79.3B.
Subsidiary-name match — cross-referenced against subsidiary lists already extracted from other companies' own annual-report exhibits, verified: 154 resolved links, $10.2B (net of overlap).
A ~$290B ceiling remains where no co-filer and no existing subsidiary record points to a parent — closing it further needs a new build (guarantor lists from bond prospectuses, or targeted extraction of a parent's own filings).
| Item | Status |
|---|---|
| 10-K debt-footnote extraction | The real path to a full per-issuer debt stack (bonds + term loans + private placements together, not fund-held bonds alone). Scoped, not started; the fund-held panel above will keep the maturity-wall signal live in the meantime. |
| Pre-2016 8-K backfill | An estimated ~21,000 more issuer-debt events sit in 2001–2015 filings; scoped and ready the moment the underlying search index extends that far back. |
| Remaining parent-linkage ceiling | ~$290B with no resolved parent yet — a scoping decision on whether a dedicated extraction pass is worth building next. |
Numbers live-verified against primary tables on 2026-09-24. Every row in every table above
traces to a specific SEC accession number or filing date — source_provenance and
integrity_tier travel with the data, not just the summary. Prepared for a live
client conversation; ask for the underlying query if a number needs to be reproduced live.