Working with SEC EDGAR, written by people who run it nightly.
Each of these is a thing we got wrong first. The code matches what our own pipeline does, and the traps are the ones that cost us a night.
Parse Form D with Python
Find the filings, read the XML, and avoid the four fields that quietly give you a wrong number.
Read it →Read 8-K item codes
Classify an 8-K from its own header, and cap the download before it costs you 226 MB of memory.
Read it →EDGAR rate limits and access
One budget across every sec.gov host, a User-Agent that names you, and the retry rule that keeps you out of trouble.
Read it →Map a ticker to a CIK
One public file, a many-to-one relation, and the reason CIK is the identifier you store.
Read it →EDGAR is free. The pipeline around it is the cost.
Fetching a filing is an afternoon. Running a daily pass that survives backdated arrivals, unbounded submission files, a filer inventing a new item title and a weekend with no index is the part that keeps going. Build it, or read ours as JSON.
If you already know what you want to detect rather than how to parse it, the use-case pages start from the job.
Browse the use cases →