Skip to content
ko.io

Data reference

Every dataset, traced to its source

What each dataset contains, the filing or release it is built from, how far back it goes, how often it updates, and the REST route and MCP tool that serve it. Select a dataset to inspect it and fetch a live record.
Get API keyAPI reference

  • 23

    datasets

  • 100M+

    source-traced rows

  • 1987

    earliest 13F quarter on file

  • 2026-06-03

    row counts measured

Start with the core datasets

Ownership, trading and macro series that most research starts from. The full directory with every field and route follows below.

13F holdings

Quarterly positions for 12,900+ institutions, one row per issuer, consolidated across filing entities.

Quarterly13F-HR

Form 4 insider

Officer, director and 10% owner transactions, open-market buys and sells separated from awards.

DailyForm 4

Congress PTR

House and Senate periodic transaction reports with the statutory amount range.

As disclosedSTOCK Act

Fails-to-deliver

SEC fails-to-deliver by settlement date, keyed to ticker and CUSIP.

Twice monthlySEC FTD

Treasury yields

The daily Treasury par yield curve across every tenor, since 1990.

DailyU.S. Treasury

8-K events

Material events such as buyback authorizations, extracted from current reports.

Event-driven8-K

Directory

23 datasets

Ownership & insider

6

Who owns what, and who is buying or selling their own stock — from SEC ownership filings.

Corporate & governance

4

Material events, fundamentals, activist stakes and executive pay.

Market structure & government

4

Settlement failures, threshold lists, and the trades of members of Congress.

Macro & economic

9

Rates, the curve, the Fed balance sheet, inflation, positioning, banks and energy.

Row counts and coverage windows measured against the production cluster on 2026-06-03. Live samples are real responses from https://api.ko.io; signed-out requests use demo mode, and some endpoints need a paid plan.

Method

How the data is built

A row is only published if it can be traced back to the filing or release it was parsed from.

  1. Source
    EDGAR filing
    The original SEC document
  2. Identifier
    Form, period, filer
    Which filing a row came from
  3. Structure
    Parsed row
    Normalised, linked fields
  4. Delivery
    Web · REST · MCP
    Same row everywhere
Every row names the form, period and filer it was parsed from.
Source-traced
Every SEC row carries the accession number of the filing it came from, so any figure can be opened in the original document. Nothing is editorial.
Normalized
Share classes (GOOGL + GOOG) are consolidated, and filing entities that restructure are merged back to one canonical manager so portfolios stay continuous.
Tested before publish
Uniqueness, not-null, range and reconciliation tests run on every build. For the 13F serving tables a failed test blocks the publish, so the API keeps serving the last good version.
Replicated
Stored on a three-replica database cluster. The terminal, the REST API and the MCP server all read the same tables.

Query it yourself

Every dataset is one REST call away; most have an MCP tool. Start on the free tier.

Get API keyConnect MCP