Custom Data Extraction from SEC Filings: AI-Powered Solutions

Most data providers give you a standard set of fields. That covers most needs, but some research depends on a specific clause or term that no standard dataset tracks. That's where custom data extraction comes in.

Prefer video? Watch the full walkthrough.

What standard extraction already covers

Ask Edgar uses fine-tuned AI models to turn SEC filings into structured data, with people checking the results. The standard dilution data includes:

• pending shelf registrations and ATM programs,
• outstanding warrants and convertible securities,
• completed offerings and their terms,
• exercise prices and expiration dates,
• the banks and underwriters involved,
• price protection terms, and
• registration status.

For a warrant, for example, that means whether it's registered, how many are outstanding, the exercise price, the banks involved, any price protection, and the dates it becomes exercisable and expires.

When standard data isn't enough

Some strategies need data points that exist in the filings but aren't part of any standard dataset:

• Specific clauses. A fund might want every offering that combines a right of first refusal with a tail financing provision. The offering is tracked, but that combination of clauses isn't flagged.
• Custom metrics. How long a right lasts, specific underwriter fee structures, or which agents replaced which.
• Historical research. Backtests often need the same data pulled from years of older filings.

How custom extraction works

1. Define the data points. We work with you to pin down exactly which clauses, numbers, dates and relationships you need.
2. Backfill history. We run the extraction over past filings, going back as far as you need.
3. Keep it current. The same extraction runs on new filings every day.
4. Deliver it your way: daily Excel exports, through the API, or as custom fields on an enterprise dashboard, alongside the standard dilution data.

Once the custom data sits next to the standard data, you can filter on both together. Using the right of first refusal example, you could look for offerings with a particular clause combination, deals nearing expiration, particular underwriters, or a fee structure.

Things to know

• Accuracy: complex legal language can need extra checking for custom fields.
• Older filings: formatting and wording change over time, which can affect how complete the historical data is.
• Upkeep: extraction rules may need updating as deal terms and regulations change.

If your research depends on data points no standard feed covers, get in touch through the Ask Edgar site to talk through what you need.