FinScraps

Open-source scrapers for Brazilian financial-market datasets.

ANBIMA IRTS Dataset

FinScraps currently publishes ANBIMA Interest Rate Term Structure (IRTS) parameters as a public Parquet dataset.

• Dataset repo: https://huggingface.co/datasets/rodrigomtorresb/anbima-irts
• File: latest.parquet
• Schema: date, type, b1, b2, b3, b4, l1, l2

The dataset is accumulated in a single latest blob. Generated data is published to Hugging Face and is not committed back into this Git repository.

ANBIMA IDKA Dataset

FinScraps also publishes ANBIMA IDKA historical index levels as a wide public Parquet dataset.

• Dataset repo: https://huggingface.co/datasets/rodrigomtorresb/anbima-idka
• File: latest.parquet
• Schema: date, IDKAPRE3M, IDKAPRE1A, IDKAPRE2A, IDKAPRE3A, IDKAPRE5A, IDKAIPCA2A, IDKAIPCA3A, IDKAIPCA5A, IDKAIPCA10A, IDKAIPCA15A, IDKAIPCA20A, IDKAIPCA30A

Each IDKA column contains ANBIMA's Número Índice value. Derived return columns from the source workbooks are intentionally omitted.

Usage

import pandas as pd

url = "https://huggingface.co/datasets/rodrigomtorresb/anbima-irts/resolve/main/latest.parquet"
df = pd.read_parquet(url)

import pandas as pd

url = "https://huggingface.co/datasets/rodrigomtorresb/anbima-idka/resolve/main/latest.parquet"
df = pd.read_parquet(url)

Automation

The scraper runs on GitHub Actions after Brazilian market close. It fetches the previous Brazilian business day for IRTS, rebuilds the cumulative IDKA wide frame from ANBIMA's historical workbooks, validates the resulting datasets, and uploads replacement files to Hugging Face when data changed.