Radar Perene / Archive / science

B3 open data: what is public and what is paid

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

Science

The quote that shows up for free on a news portal arrives late — usually fifteen minutes after the trade happened. The investor who wants the price in the instant pays for it. That boundary, which every home-broker user knows by sight, organizes the entire world of Brazilian exchange data: real time is the product; the historical record is, to a large extent, public patrimony. And for research — contrary to what people assume — the free side of the boundary is the more valuable one.

B3 makes the closing-price history of the equity market available for free — the daily series of prices, volumes and trades — while charging for real-time data and for the analytical products built on top of them. The distinction is not one of quality but of latency: the same trade that costs money to watch as it occurs becomes public at the day's close.

The public side: deeper than it looks

The centerpiece of the free archive is the historical quote series — the files known by the technical name COTAHIST, published on B3's website. In numbers: the annual files cover the market since 1986, security by security, session by session — four decades of closes, highs, lows and volumes, downloadable with no registration. The format is spartan (fixed-width files with a documented layout), which tends to frighten on first contact and stop mattering on the second: once parsed, the file is among the most complete datasets a market researcher will find for free anywhere in the world.

Around that centerpiece there is more: the compositions and methodologies of the indices (Ibovespa, IFIX, the sector indices), the daily bulletins, market statistics and the developer portal with query APIs. For research questions — regimes, cycles, cross-sector relationships, long-horizon behavior — none of this costs a cent.

One professional caveat comes with the gift, though: the prices in the historical files are traded prices, unadjusted for corporate actions. Dividends, splits and reverse splits appear as steps in the raw series, and adjusting for them is the researcher's task — the exchange provides the faithful record of the session, not a total-return-ready series. Whoever compares a study built on adjusted prices with one built on raw prices is, without noticing, comparing two different objects. It is the kind of detail no honest methodology summary can omit.

The subscribed side: for those to whom time matters

What the exchange sells is, essentially, speed and depth. Real-time market data feeds, distributed to brokers and vendors, power screens and algorithms that need the price now; the full order book and detailed intraday data follow the same regime. Packaged analytical products, such as the exchange's own data platforms, are likewise contracted. It is a division consistent with who uses what: trading needs the instant; research needs the history.

What the house does with the free side

The essays in Radar Perene's archive — the regime readings, the studies of index ratios, the historical episodes of the Atlas — are built on closing data, precisely the public layer described above. A published example: the study of the IFIX/IBOV ratio as a risk-aversion thermometer in March 2020, which the reader will find in "The IFIX/IBOV as an aversion thermometer", uses daily index series only — nothing a patient reader could not rebuild from the same source. The house's working papers on Brazilian intramarket relationships, published with a DOI on Zenodo (10.5281/zenodo.21327663), follow the same diet: daily closes, public source, declared method.

There is a lesson of method embedded in that choice, and it deserves a sentence: research that only works with expensive data is research almost nobody can check. B3's free layer imposes a limit — no microstructure, no intraday — and that limit is also a protection, because it keeps every result within reach of someone else's verification.

A trap in the history

The public archive carries the companies that left the exchange — and that is exactly why it is worth so much. The classic mistake runs the other way: building a historical study from today's list of companies, unknowingly inheriting only the survivors. COTAHIST makes the mistake avoidable; it does not make it impossible. The difference between those two things is the subject of another article in this section, on survivorship bias.

Frequently asked questions

Where are B3's historical files?

In the market data and indices area of B3's website, under "historical series." The download is open and the file layout is documented on the same page.

Are the free data good enough for backtesting?

They are good for studies on closing data — which is the horizon of most research questions. Intraday strategies would require subscribed data, and age on top of a different infrastructure.

Is B3's permission needed to research with the public data?

For research and analysis, the open files are freely accessible. Commercial redistribution of data, above all in real time, is what requires a contract.

How do B3's data differ from the Central Bank's SGS?

They complement each other: the SGS covers aggregate macroeconomic and financial series; B3 covers the market security by security. Much of the house's work crosses the two.

---

Continue the track: What ORCID is and why independent researchers need one too →

House readings: today's reading is in the Diário; the episodes rebuilt on those files, in the Atlas.

Turning four decades of public files into a curated research base is backstage craft — the kind the house details in conversation, not in an article.