SEC Form 13F: Institutional Investment Manager Holdings Disclosure β Engineer's Guide for 2026
Building FinanceTrackDaily on the SEC EDGAR API, I read a lot of filings that retail investors obsess over β 10-K annual reports, 8-K material events, Form 4 insider trades. But no filing generates more breathless headlines than Form 13F. Every quarter, financial media runs the same story: "Here is what Warren Buffett bought." "The hedge fund that loaded up on this chip stock." Almost all of that coverage traces back to a single, dry quarterly disclosure that institutional managers are legally required to file. From an engineering perspective, Form 13F is one of the most structured and queryable datasets the SEC produces β and also one of the most misunderstood. This is what aggregating it across thousands of managers taught me about what it really contains, and what it quietly leaves out.
What Form 13F Actually Is
Form 13F is a quarterly report of equity holdings that certain large investment managers must file with the SEC. The requirement comes from Section 13(f) of the Securities Exchange Act of 1934, added by Congress in 1975 to give regulators and the public visibility into the holdings of large institutional money managers. The triggering threshold is exercising investment discretion over at least 100 million dollars in "Section 13(f) securities" β a defined universe that is mostly exchange-traded U.S. stocks, certain equity options, ETFs, and convertible debt.
One detail I find genuinely striking as someone who ingests this data: the 100 million dollar threshold was set in 1978 and has never been adjusted for inflation. In 1978 dollars, 100 million was a serious institution. Today, after decades of inflation, far more managers cross that line than Congress likely envisioned, which is one reason the 13F universe has grown so large. The SEC proposed raising the threshold to 3.5 billion dollars in 2020, but that proposal was not adopted, so the 1978 figure still governs in 2026. That is the kind of regulatory-history fact that matters when you are sizing how many filers your aggregator has to handle.
The SEC itself publishes the "Official List of Section 13(f) Securities" every quarter. If you are building anything on 13F, that list is your authoritative universe β a security only needs to be reported if it appears on it. I treat that quarterly list as a join key: holdings referencing a CUSIP that is not on the current list are, by definition, outside the reporting requirement.
The Two Flavors: 13F-HR and 13F-NT
A practical branching point for any parser is that "13F" is not one form type in EDGAR. The two you will encounter most are 13F-HR (the Holdings Report, which carries the actual information table of positions) and 13F-NT (the Notice, filed when another manager is reporting the holdings on the filer's behalf). There are also combination and amendment variants (13F-HR/A, 13F-NT/A). If you ingest only by the string "13F" you will pull in notices that contain no holdings and wonder why your position table is empty for some filers. Branch on the exact form type early.
The Filing Schedule and the 45-Day Lag
Form 13F is due within 45 days after the end of each calendar quarter. So Q4 holdings (as of December 31) are due by roughly February 14; Q1 holdings (as of March 31) by mid-May, and so on. This 45-day window is the single most important thing to understand about 13F data, and the thing most retail headlines gloss over.
By the time a 13F is public, the holdings it describes can be up to 45 days stale β and that is the snapshot as of quarter-end, which may itself be weeks older than the filing date. A manager who held a stock on December 31 may have sold the entire position in January, well before the February filing reveals it. When a headline says "this fund just bought," what actually happened is "this fund held, as of a quarter-end date that is now six-plus weeks in the past." Building a system that timestamps both the period-of-report date and the filing date separately is the only honest way to represent this.
The Data Shape: The Information Table
The heart of a 13F-HR is the information table, which EDGAR exposes as a structured XML document. Each holding is an infoTable element, and the fields are consistent and machine-friendly. The core ones I map are nameOfIssuer, titleOfClass, cusip, value, and inside shrsOrPrnAmt the sshPrnamt (share or principal amount) and sshPrnamtType. There is also investmentDiscretion, an optional putCall flag for options, and voting-authority columns split into sole, shared, and none.
Because it is real XML with a published schema rather than free-text PDF, 13F is a pleasure to parse relative to most SEC filings. The CUSIP field is the natural join key to map a holding to a security, and from there to a ticker. One caveat from experience: CUSIPs in 13F filings are occasionally malformed, truncated, or padded by the filer's software, so any robust pipeline needs a validation and normalization step on that field rather than trusting it blindly.

A Critical Change: Value Now Reported in Dollars, Not Thousands
Here is a data point that has tripped up more than one analyst and that I had to handle explicitly in ingestion. For decades, the value field on Form 13F was reported in thousands of dollars β a position worth 5 million dollars was entered as 5,000. The SEC amended Form 13F, and effective for filings on or after January 3, 2023, the value column is reported in actual dollars (ones), not thousands.
This means any historical 13F dataset has a discontinuity: pre-2023 values are in thousands, post-2023 values are in ones. If you naively concatenate them, a manager's reported portfolio appears to balloon a thousandfold overnight. An aggregator that spans the boundary must scale older records or store the reporting unit alongside each value. This is exactly the kind of silent schema change that breaks dashboards months later, so I flag the period-of-report date against the January 2023 cutoff during ingestion.
What Form 13F Does Not Tell You
The biggest mistake retail readers make is treating a 13F as a complete picture of a fund. It is not, by design, and the gaps are large:
No short positions. Form 13F reports long positions in 13(f) securities only. A fund could be net bearish on a name it appears to "own," and you would never see the offsetting short. This single omission makes naive "copy the whales" strategies dangerous.
No cash, bonds, commodities, or foreign-listed shares. The 13(f) universe is specific. Treasuries, most fixed income, physical commodities, and shares listed only on foreign exchanges do not appear. A 13F is a slice of a portfolio, not the portfolio.
No cost basis and no intra-quarter activity. You see the end-of-quarter snapshot. A manager who bought and sold the same stock three times within the quarter shows only the final resting position β or nothing, if they were flat at quarter-end.
Confidential treatment. Managers can request confidential treatment to delay disclosure of certain positions, typically while they are still building or unwinding them. Granted requests mean some holdings are legally absent from the table you are reading until a later date. From an engineering standpoint this shows up as filings that quietly under-report, then get amended.
Why I Still Find 13F Valuable
Despite all of that, 13F is one of the richest public datasets in U.S. markets because it is broad, structured, and consistent across thousands of filers. Aggregating it, I can answer questions no single filing answers: how concentrated is institutional ownership of a given stock, which managers entered or exited a name quarter-over-quarter, and how holdings cluster across a sector. The value is in the cross-section and the time series, not in any single "the famous investor bought X" headline. Treated as a quarterly, partial, lagged census of institutional long equity positions β which is exactly what it is β it is enormously useful. Treated as a real-time trade feed, it is misleading.
Filing Mechanics and Access
All 13F filings are submitted through EDGAR and are freely available. The structured information-table XML, rather than the human-readable cover page, is what you want to ingest programmatically; EDGAR's full-text search and the per-filer filing index both expose 13F submissions by form type and date. There is no filing fee. Managers file electronically, and the same EDGAR access rules β fair-use rate limits and a declared user-agent β apply as for any other filing type. If you are pulling 13F at scale, respecting EDGAR's published request limits is not optional; it is the difference between a stable pipeline and a throttled one.
Important Disclaimer
This article is for informational and educational purposes only and is not financial advice, investment advice, legal advice, or tax advice. The author, Fanny Engriana, is a software engineer who builds public-data aggregators using the SEC EDGAR API. The author is not a registered investment adviser, broker-dealer, CFA, or CFP. Form 13F filings are historical, quarter-end snapshots of long equity positions that are reported with a delay of up to 45 days and that omit short positions, leverage, cash, and many asset classes β they do not represent a manager's complete portfolio and do not predict future performance. Nothing here is a recommendation to buy or sell any security. Securities regulations, filing thresholds, and SEC schemas change over time, so always verify the current rules against the official sec.gov source before relying on them. Consult a licensed financial advisor for personal investment decisions, a licensed attorney for legal questions, and a licensed CPA for tax questions.
Conclusion
Form 13F is the SEC's quarterly window into what large institutional managers hold β broad, structured, freely available, and genuinely well-suited to aggregation. But its value depends entirely on respecting its limits: the 45-day lag, the long-only scope, the missing shorts and cash, the confidential-treatment gaps, and the 2023 switch from thousands to dollars that silently breaks any dataset spanning it. For an engineer building on SEC EDGAR, the discipline is the same as with any filing type β record what the data actually represents, timestamp the period and the filing separately, normalize the CUSIPs, and never let a clean XML schema fool you into thinking a partial census is a complete portfolio. Read that way, 13F is one of the best public datasets U.S. markets produce. Read as a stock-tip feed, it is a trap.
Authoritative sources used throughout this article: U.S. Securities and Exchange Commission, sec.gov; Section 13(f) of the Securities Exchange Act of 1934; SEC Form 13F and Official List of Section 13(f) Securities; SEC Final Rule amending Form 13F reporting (effective January 2023); SEC EDGAR Filer Manual and Frequently Asked Questions About Form 13F.
Found this helpful?
Subscribe to our newsletter for more in-depth reviews and comparisons delivered to your inbox.