Financial knowledge graph
A financial knowledge graph is a database that stores companies, people, products, regulators, events and filings as connected entities, with the relationships between them, such as supplier, customer, lender or subsidiary, stored as links. Because the links are explicit, software can follow a change from one company to the others it affects.
How a knowledge graph differs from a table
Most financial data lives in tables: one row per company, one column per number. Tables are good at comparing companies on the same measure, but they hide how companies depend on each other. A knowledge graph stores those dependencies directly. Each company, person, product or event is a node, and each relationship is an edge with a type, a date and a source.
That structure makes multi-step questions natural. "Which of my holdings buy from a supplier that sells into a market that was just restricted?" is three hops through a graph, and very hard to answer from a spreadsheet.
How a financial knowledge graph is built
Building and maintaining one is mostly a reading problem.
- Read sources: filings, earnings calls, news, regulators, prices and macro releases.
- Resolve entities: decide that two different names refer to the same company, and keep one record for it.
- Extract relationships: supplier, customer, lender, competitor, subsidiary, regulator and the events that link them.
- Date and source every edge, so it is clear when a relationship was true and where it came from.
- Keep it current: new relationships are written in as they appear, and old ones are kept with their dates.
Use cases in finance
Knowledge graphs are used across financial services wherever relationships matter more than single numbers.
- Supply chain exposure: which companies depend on a supplier, a region or a commodity.
- Event impact: how a rule change, a rate decision or a profit warning spreads beyond the company in the headline.
- Hidden concentration: different holdings that share the same customer, lender or risk.
- Idea generation: companies the evidence points to that are not yet on a watchlist.
- Counterparty and compliance work, such as mapping ownership and control.
An example
A government proposes new export restrictions on advanced chips. A table of semiconductor stocks shows which chip makers are named. A knowledge graph goes further: it finds the equipment makers that sell into the restricted markets, the companies that buy from those equipment makers, and which of those companies are in your portfolio. One holding turns out to be exposed through its supplier, not through anything it sells itself.
Illustrative example, based on the research run shown on the Clargus research agent page.
How Clargus uses a financial knowledge graph
Clargus keeps a knowledge graph of companies, suppliers, customers, regulators, events and filings. Sources are read around the clock, new companies and relationships are written in dated and sourced, and effects are traced through suppliers, customers and lenders to the companies a change reaches. Each thesis that rests on an affected company is re-checked and marked as holding, weakened or broken, with the reason.
The graph is one of nine families of Clargus building blocks, covering entities, multi-hop exposure and influence ranking. Your own research brain sits on top of the shared graph, so market changes reach your portfolio and theses directly.