Tony Wang8 min readWe Indexed All 976,925 SEC Filers: EDGAR Is Not a List of Companies
976,925 EDGAR filers indexed: only 4.7% are operating companies, Delaware incorporates 56%, and ETFs dominate the 13F ownership graph.
Ask most people what's in SEC EDGAR and you'll get some version of "public company filings." That's true in the sense that a library contains Shakespeare — technically correct, wildly unrepresentative of the shelves.
We indexed the whole filer registry — every CIK, 976,925 of them, one record each, built from SEC's own submissions feeds, XBRL company facts and 13F holdings — and the shape of it is not what the phrase "SEC companies" suggests. Fewer than 1 in 20 filers is an operating company. The rest is a long tail that nobody puts on a landing page.
94.9% of EDGAR is not a company
One record per CIK, bucketed by the entity type EDGAR itself reports:
That 927,522 is not junk data — it's the actual system. Every corporate officer and director who files a Form 3 when they take a role, and a Form 4 every time they trade, gets a CIK. Every startup that raises a private round and files a Form D gets a CIK. Every mortgage securitization trust gets a CIK. They are all "SEC filers." Almost none of them are what you'd screen for.
This matters practically: if you query an SEC dataset for "companies" without filtering, you're mostly querying people. Filter entity_type=operating and the 977K becomes 45K.
Delaware wins by more than everyone else combined
Of the 582,315 filers that record a state or country of incorporation, Delaware accounts for 327,228 — 56.2%, which is to say more than every other jurisdiction on Earth put together.
The Delaware effect is well known as a talking point. It's another thing to see it as a raw count with no survey methodology in between — it's simply what filers write on their own forms.
The most-filed form in EDGAR is Form D
Not the 10-K. Not the 8-K. The private-placement notice.
Only 39,399 filers have ever filed a 10-K — about 4% of the registry, and fewer than the number who have filed a Form D nine times over. The mental model of EDGAR as "where public companies file annual reports" describes a real but small corner of it. By filer count, EDGAR is mostly private fundraising and Section 16 insider disclosure.
The industry mix says the same thing from another angle. The largest classified industry is Asset-Backed Securities (12,546 filers) — securitization trusts — followed by Blank Checks (3,313), the SPAC/shell category. Pharmaceutical Preparations (1,899) and Prepackaged Software (1,773) trail both.
ETFs ate the ownership graph
The 13F side is a separate index: 1,748,475 position lines reported by 9,445 institutional managers on their latest 13F-HR. Counting how many managers report a position in each issuer:
Six of the top seven are fund families, not companies. iShares alone shows up in 95,989 position lines — ten times the most widely-held individual operating company, Alphabet, at 9,160. Institutional "ownership" as disclosed on 13F is, in bulk, ownership of wrappers.
The trap: EDGAR's biggest company is Toyota (if you don't read the units)
This one cost us a bug, so it's worth stating plainly for anyone building on XBRL.
Sort the dataset by reported annual revenue, descending, and the top ten comes back entirely Japanese — Toyota, NTT, Mitsubishi UFJ, Hitachi, Sony, Mizuho. Toyota's reported figure is 30,225,681,000,000. Read carelessly, that's a $30 trillion company, roughly a third of world GDP.
It's ¥30.2 trillion — about $200B. XBRL company facts store each figure under a unit key that is the filer's native reporting currency, and SEC does not normalize it. A parser that grabs the value and assumes dollars will rank every yen-, won- and rupiah-reporting filer above every US mega-cap, forever, and produce a "largest companies" list that is really a "largest currency denominations" list.
The fix is unglamorous: read the unit key as the reporting currency, keep the native figure exactly as filed, and store a separate converted field for anything you intend to sort, band or range-filter on. Our dataset now exposes reporting_currency alongside both the native latest_annual_revenue and a normalized latest_annual_revenue_usd, and every money sort and filter runs on the normalized field.
The related trap: foreign private issuers file under the ifrs-full taxonomy, not us-gaap. A parser that reads only us-gaap concepts silently returns zero financials for them — not an error, just an empty statement. Coverage looks fine until you notice which companies are missing.
What it's for
Screening the whole registry rather than pulling one filing at a time: every filer by industry, jurisdiction, revenue band or filing activity; one company's normalized statement history and insider trades; institutional ownership from the manager side. All of it from SEC's free public data — the work is in the cleaning, the tag normalization, and the units.
The dataset is at /datasets/sec and queryable over one REST API. If you take one thing from the numbers above, make it this: filter entity_type before you count anything.
Frequently asked questions
How many companies file with the SEC?
976,925 filers have a CIK in EDGAR, but only 45,497 (4.7%) are operating companies and 3,906 are investment entities. The other 927,522 are individual Section 16 insiders, one-time Form D issuers, blank-check shells and securitization trusts. If you want companies specifically, filter entity_type=operating rather than counting the registry.
What percentage of companies are incorporated in Delaware?
Of the 582,315 EDGAR filers that record a state or country of incorporation, 327,228 - 56.2% - are incorporated in Delaware. That's more than every other US state and foreign jurisdiction combined; the next-largest US state is Texas at 22,325 (3.8%), then California at 17,388 (3.0%).
What is the most commonly filed SEC form?
Form D, the private-placement exemption notice, at 357,255 filers - ahead of Form 4 (281,810) and Form 3 (281,670), both insider-disclosure forms. Only 39,399 filers have ever filed a 10-K. By filer count EDGAR is primarily a private-fundraising and insider-disclosure system, not a public-company reporting one.
What stock do institutional investors hold most?
Across 1,748,475 13F positions from 9,445 managers, the most widely-reported issuer is iShares Trust at 95,989 position lines - an ETF family, not a company. Six of the top seven are fund families. The most-held individual operating company, Alphabet, appears in 9,160 - roughly ten times fewer than iShares.
Why does SEC revenue data show Toyota as the largest company?
Because XBRL company facts store each figure under a unit key that is the filer's native reporting currency, un-normalized. Toyota reports 30,225,681,000,000 - that's ¥30.2 trillion (about $200B), not $30T. A parser that reads the value and assumes dollars ranks every yen- or won-reporting filer above every US mega-cap. You have to read the unit key as the currency and convert before sorting or range-filtering.