Brazil CVM Filings for Quant Research Pipelines

You are a quant or data engineer building a fundamental layer for Brazil. You need a Brazil filings quant API: ticker → annual/quarterly PDFs → machine-readable text — not another price feed.

apicvm is built for that regulatory slice: resolve issuers, list DFP/ITR/FRE, download originals, and extract page-level markdown asynchronously.

The persona

Cross-asset quants adding B3 names beside US EDGAR pipelines
Research shops assembling panel datasets from audited statements
Feature stores that cite original filings, not scraped HTML

You already have prices elsewhere (B3 vendors, brokers). The gap is programmatic CVM documents.

The problem

Open-data ZIP/CSV dumps are excellent for some panels but awkward for “fetch VALE3 DFP 2024 PDF, then next quarter’s ITR.” Scrapers break. Multi-market APIs may bundle Brazil with event feeds you do not need yet.

Quant code wants a boring HTTP loop:

for ticker in watchlist:
  resolve → list DFP/ITR by year → download or extract → store with document.id

Solution flow

GET /v1/companies/resolve — confirm issuer
GET /v1/documents — filter type, year, ticker
GET /v1/documents/:id/file — original PDF
Optional: POST /v1/document-text-extractions — page markdown via callback

export APICVM_KEY='apicvm_...'
export APICVM_URL='https://apicvm.dev'

curl -H "Authorization: Bearer $APICVM_KEY" \
  "$APICVM_URL/v1/documents?ticker=VALE3&type=DFP&year=2024&perPage=20"

Python sketch — multi-ticker DFP index

import os, requests

BASE = os.environ["APICVM_URL"]
H = {"Authorization": f"Bearer {os.environ['APICVM_KEY']}"}
TICKERS = ["PETR4", "VALE3", "ABEV3"]

index = []
for t in TICKERS:
    r = requests.get(
        f"{BASE}/v1/documents",
        headers=H,
        params={"ticker": t, "type": "DFP", "year": 2024, "perPage": 50},
    )
    r.raise_for_status()
    for d in r.json()["data"]:
        index.append({"ticker": t, "id": d["id"], "name": d["name"], "dateRef": d["dateRef"]})

print(len(index), "DFP files indexed")

Respect rate limits when the watchlist grows — 60 requests / 60 seconds on standard plans.

What this is not

Not market data — no quotes, order books, or corporate-action calendars
Not real-time CVM firehose — corpus follows ingestion, not a live portal mirror
Not IPE / fato relevante alerts — see IPE

Next steps

Ready to integrate?

Get an API key and start querying Brazilian CVM filings programmatically.

Read the API docs Get API key