Radar Perene / Archive / science

CVM open data: what the regulator publishes beyond enforcement

◦ Index methodology v2.2 (working papers with DOI). See the methodology.

Science

The CVM appears in the news almost always in the sheriff's role: a ruling, a fine, a new rule, a formal notice. The role exists and matters — but it hides a second trade of the agency that directly interests whoever researches: the CVM is, perhaps, the largest publisher of microdata on the Brazilian capital market. The material it forces the market to report does not stay locked in an examiner's file; it returns to the public, structured, on an open data portal most independent researchers have yet to discover.

The CVM's open data portal makes freely downloadable the documents and filings that regulated participants are required to submit: daily and registration filings of investment funds, financial statements and forms of listed companies, public offering data and the general registry of market participants.

The logic that generates the archive deserves a paragraph, because it explains its quality. These data were not collected out of statistical kindness: they are byproducts of regulation. The fund reports its quota because the rule requires it; the company delivers its balance sheet because the law commands it. The data are therefore born with three rare properties — mandatory (near-complete coverage of the regulated universe), standardized (everyone reports in the same format) and dated (delay is an infraction). Empirical research lives on exactly those three properties.

The portal's two treasures

For market research, two collections stand out. The first is the investment fund filings: quota, net assets, inflows and redemptions, fund by fund, day by day — the history of the entire industry, including the funds that closed. The methodological caveat is immediate, and it is the one treated in this section's article on survivorship bias: studying the fund industry only through the vehicles alive today is studying the winners; the CVM archive lets one rebuild the cemetery, and the cemetery is half the statistic.

The second treasure is the listed-company documents: annual and quarterly statements, registration and reference forms, material fact notices — the raw material of any serious fundamentals study in Brazil, at the primary source, with no commercial intermediary. Whoever has paid for packaged balance-sheet data discovers, with some melancholy, that its origin was public.

The same body at both ends: the data and the risk

There is a symmetry in this portal that the house's experience records up close. In the regulatory monitoring that feeds Radar's wealth-intelligence line, the CVM is classified as a source of exogenous vulnerability — the vector that groups the institutions whose decisions change the rules of the game from the outside in: a new rule, a revised understanding, a ruling that changes, at a stroke, the value of an entire structure. The same body therefore occupies both ends of the research table: it is the risk variable one monitors and the data repository with which one builds the test. That is not a paradox; it is the normal anatomy of a regulator that also publishes — and the researcher who understands both functions reads both the data and the rules better.

That double role has a practical consequence for study design: the CVM's own rule changes are dated, public, documented events — natural quasi-experiments. The Brazilian literature has explored windows of this kind (changes in the information regime, new disclosure obligations) precisely because the rule's date is known with official-gazette precision. The portal supplies the data; the normative history supplies the temporal cut. Both halves of the design come from the same house.

Honest limits of the archive

The portal delivers raw files, not a ready-made database: long series require joining files by period, harmonizing layout changes and handling restatements — companies correct documents, and the researcher decides, and declares, whether to use the original or the restated version. None of that is a defect; it is the normal physiognomy of administrative data. But it explains why "the data are public" and "the study is easy" are different sentences.

Frequently asked questions

What exactly is there about funds on the CVM portal?

Daily filings (quota, net assets, flows), registration filings, portfolio compositions with a regulatory lag and accountability documents — for the universe of regulated funds, including those already wound down.

Do CVM data overlap with B3 data?

They complement each other: B3 publishes what happened in the trading session (prices, volumes); the CVM publishes what participants report (balance sheets, quotas, portfolios, offerings). Fundamentals or fund studies typically use both sources.

Do I need authorization to use them in research?

The open data portal's datasets are free to access and use. Good citation practice — dataset, consultation date — applies as with every source on this trail.

Why do some data appear with a delay?

Part of the filings carries a lag provided for in the rules — portfolio compositions, for example — to protect strategies in progress. The delay is a published rule, not a portal failure: it enters the study design like any other declared constraint.

---

Continue the trail: one house remains, the one that gathers everything — more than eleven thousand series under the same address: Ipeadata →

House reading: today's reading is in the Diário; the episodes the regulatory data illuminate, in the Atlas.

Building a fund or company database for a specific question is backstage craft — the kind the house executes on commission.