Brazil Financial Restatement Disclosures API: Track Retifications and Reissued DFP/ITR from CVM Filings

The problem

Restatement surveillance splits into three stacks that English search results often merge:

1. Market data terminals — may flag “accounting changes” from third-party feeds. They rarely store the issuer’s regulatory PDF explaining what was retified, which periods moved, and whether the change is presentation-only vs error correction.
2. Periodic packages — DFP and ITR rows that get replaced after the first publication. You need stable document.id tracking per dateRef, not just “latest year=2024” without noticing a superseding upload.
3. Event disclosures — fatos relevantes and market communications when management retifies prior quarters, reclassifies revenue, or corrects control failures outside the normal earnings calendar.

Scraping the CVM portal works for one ticker and breaks on a diversified watchlist. Bulk open-data ZIPs are dataset-oriented, not ticker-first. You want GET /v1/documents with stable document.id, then GET /v1/documents/:id/file for the original PDF — scoped to restatement and retification workflows.

For assurance language that often precedes a retification, cross-check Brazil independent auditor report disclosures API. For revenue-policy changes that drive reclassifications, see Brazil revenue recognition disclosures API. This guide focuses on the API list-and-download pattern for retification events and reissued periodic packages.

Which filing type carries restatements?

Brazilian issuers do not expose a dedicated type=RESTATEMENT in the apicvm document list. Restatement detail usually appears under:

Practical rule: poll FATO_RELEVANTE and COMUNICADO_AO_MERCADO with Portuguese and English keywords (retific, reclass, restate, correção, período anterior), track document.id per ticker and dateRef on ITR/DFP, and pull FRE once per year for control-failure context. After download, search PDF text for retificação, reclassificação, reapresentação, períodos anteriores, restatement, correção de erro, or ajuste retrospectivo.

Issuer packaging varies: some filers publish a short fact PDF pointing to a full reissued ITR; others embed the entire corrected financial statements in the event filing. Zero rows with a keyword filter does not mean no restatement occurred — broaden the list or inspect the latest ITR/DFP for duplicate dateRef with different id.

How apicvm helps

1. Resolve the issuer — GET /v1/companies/resolve
2. List event filings — GET /v1/documents with types=FATO_RELEVANTE,COMUNICADO_AO_MERCADO
3. List periodic packages — GET /v1/documents with type=ITR or type=DFP and year
4. Narrow with name tokens — Retific, Reclass, Corre, or Restat
5. Scope with dateRef; sort with field=dateRef&order=desc
6. Download the PDF — GET /v1/documents/:id/file (no extraction credits)
7. Optional (Pro): page-level markdown — POST /v1/document-text-extractions with async callback

Auth: Authorization: Bearer or X-API-Key. Try VALE3 without a key via the demo API.

Setup

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

Step 1: List restatement-related material facts

curl -s -H "Authorization: Bearer $APICVM_KEY" \
  "$APICVM_URL/v1/documents?ticker=PETR4&types=FATO_RELEVANTE,COMUNICADO_AO_MERCADO&year=2025&perPage=50&field=dateRef&order=desc" \
  | jq '.data[] | select(.name | test("retific|reclass|restate|correção|correcao|período anterior|periodo anterior"; "i")) | {id, type, dateRef, name}'

Response rows include stable id, dateRef, name, and nested company metadata. Use id as your cache key — not name, which can be long and encoding-sensitive.

Paginate with page= when meta.lastPage > 1. perPage maxes at 50.

Step 2: List quarterly filings to catch reissued ITR packages

curl -H "Authorization: Bearer $APICVM_KEY" \
  "$APICVM_URL/v1/documents?ticker=VALE3&type=ITR&year=2025&perPage=50&field=dateRef&order=desc"

If two rows share the same dateRef but different id, treat the newer upload as a candidate reissue and download both PDFs for manual comparison.

Step 3: Filter annex rows that mention retification in the title

curl -s -H "Authorization: Bearer $APICVM_KEY" \
  "$APICVM_URL/v1/documents?ticker=ITUB4&type=DFP&year=2024&perPage=50&name=Retific" \
  | jq '.data[] | {id, type, dateRef, name}'

If that returns few rows, retry with name=Reclass, name=Corre, or list without name and inspect titles manually.

On the public demo (VALE3 only), swap the host path to /v1/demo/documents with the same query parameters for smoke tests.

Step 4: Download the retification PDF

DOC_ID=123456
curl -H "Authorization: Bearer $APICVM_KEY" \
  "$APICVM_URL/v1/documents/$DOC_ID/file" -o restatement-notice.pdf

Store document.id in your warehouse. Re-download when the issuer replaces a filing; do not assume name alone is a stable key.

Step 5: Watchlist poll (Python sketch)

import os, re, requests

BASE = os.environ["APICVM_URL"]
KEY = os.environ["APICVM_KEY"]
HEADERS = {"Authorization": f"Bearer {KEY}"}
WATCH = ["PETR4", "VALE3", "ITUB4", "WEGE3"]
RESTATE_RE = re.compile(
    r"retific|reclass|restate|corre[cç][aã]o|per[ií]odo anterior|"
    r"reapresenta|ajuste retrospectivo",
    re.I,
)
seen = set()

def list_events(ticker):
    r = requests.get(
        f"{BASE}/v1/documents",
        headers=HEADERS,
        params={
            "ticker": ticker,
            "types": "FATO_RELEVANTE,COMUNICADO_AO_MERCADO",
            "year": 2025,
            "perPage": 50,
            "field": "dateRef",
            "order": "desc",
        },
        timeout=60,
    )
    r.raise_for_status()
    return r.json().get("data") or []

for ticker in WATCH:
    for doc in list_events(ticker):
        if doc["id"] in seen:
            continue
        if not RESTATE_RE.search(doc.get("name") or ""):
            continue
        seen.add(doc["id"])
        print(f"RESTATE {ticker} {doc['dateRef']} {doc['name'][:70]}")

Wire this into GitHub Actions CVM filings monitor or extend the diff pattern from Brazil material facts monitoring API. For NLP over downloaded PDFs, see extract text from CVM PDFs for AI agents.

Where restatements sit in the filings stack

Brazil regulatory filings API explains the full layer map. Pair event polling with quarterly ITR monitoring during 3Q ITR season (typically Oct–Nov). When retifications follow qualified opinions, cross-check Brazil independent auditor report disclosures API. When management revises forward numbers without retifying prior periods, see Brazil guidance outlook disclosures API. For the broader Brazil financial data API pillar, treat restatements as the correction layer on top of periodic statements.

Current limitations

• Documents, not parsed restatement analytics — apicvm does not return before/after EPS, automatic line-item diffs, or “materiality” scores.
• Title heuristics — name filters are imperfect; some retifications use generic titles.
• Reissue detection — comparing duplicate dateRef rows requires your own logic; the API does not flag “superseded” explicitly.
• Polling, not push — no webhook when CVM publishes a new fact. Schedule polls and diff document.id.
• Ingestion lag — sync is not real-time. A notice may appear on the portal before it is in the corpus.
• Corpus coverage — not every issuer or period is present. Validate against the portal for production compliance use cases.
• Demo scope — /v1/demo/* is limited to VALE3. Other tickers need a production key.
• ITR disambiguation — Informações Trimestrais (CVM quarterly financials), not US income tax return software.

Next steps

• Read the API docs
• Get an API key
• Brazil material facts monitoring API — generic event poll loop

Related guides

• Brazil independent auditor report disclosures API — opinions that often precede retifications
• Brazil income statement disclosures API — DRE lines affected by reclassifications
• Brazil revenue recognition disclosures API — policy changes behind revenue restatements
• Brazil guidance outlook disclosures API — forward revisions vs backward retifications
• Track Brazil earnings season with the CVM filings API — seasonal ITR workflow
• Extract text from CVM PDFs for AI agents — async markdown extraction