Hi everyone,
Some update on PibouFilings. It is a Python library I built and maintain for pulling and parsing SEC filings (insider trades + fund holdings) from 1990 to today, in SQL, with a single function call.
I've personally used it to understand who I am trading against. There are clear patterns of stock volatility based on who is/are the market makers for a stock.
What's new in 0.5.1:
- DuckDB is the default backend now.
Parsed data lands in a single DuckDB file, one table per dataset, PK-based dedup. Easy to query, fast on tens of millions of holdings rows, no server to run. CSV export is still there if you want it (`export_format="csv"`).
- Crash-safe resume.
If a run dies mid-download, rerunning skips what's already on disk (both parsed rows and cached raw filings). No more starting over.
- Form coverage.
13F-HR (institutional holdings), NPORT-P (fund holdings), and Section 16 (Forms 3/4/5 for insider trades).
- Parallel workers
Auto-bucketed by form type (quarterly for 13F, monthly for NPORT and Section 16).
- Transparent parsers
You can keep the raw `.txt` filings and post-process them yourself if you don't trust my parsing (create a PR and update the filers ;).
Try it
Install: pip install -U piboufilings
from piboufilings import get_filings
USER_AGENT_EMAIL = "yourname@example.com" # required by SEC fair-access policy
USER_NAME = "Your Name or Company"
get_filings(
user_name=USER_NAME,
user_agent_email=USER_AGENT_EMAIL,
cik="0001067983", # Berkshire Hathaway; pass None to get all
form_type=["13F-HR", "NPORT-P", "SECTION-6"],
start_year=2020,
end_year=2025,
base_dir="./my_sec_data", # parsed data
log_dir="./my_sec_logs", # operation logs
raw_data_dir="./my_sec_raw_data",# cached raw .txt filings
keep_raw_files=True, # set False to drop raw after parsing
max_workers=5,
export_format="duckdb", # "duckdb" (default) or "csv"
)