Entities, relationships, events and signals

Records are entities; lines between them are relationships; dated changes are events; readings over events are signals. Each carries a basis that says whether it is a filed fact, a computed figure or an inference.

Everything DFX publishes is labelled with what kind of statement it is, so you always know how much weight it can carry:

  • Fact: observed in a filing or announcement. The statement is the source's, not DFX's.
  • Derived metric: computed from facts with a stated formula and time window (a share, a rate, a change between filings).
  • Signal: a detector's reading of events. A reading, not a fact.
  • Anomaly: a joined pattern no single source shows, published with what would prove it wrong.
  • Inference: rests on a candidate-grade link (for example two records matched on name alone) and is labelled so.

Current versus historical

DFX keeps every reading, not only the latest. A firm's figures are shown from its latest filing, and the history (filing to filing changes, owners through time, a fund's reporting by year) sits in the record's history and timeline sections. A relationship that ended is shown as ended, with its dates.

One identity across markets

When the same institution appears in two markets (an adviser that is also a family office vehicle's manager, a lender that is also a private equity firm) DFX joins them and says on what basis: same CRD, same SEC CIK, same EIN, same LEI, confirmed by an analyst, or same name only (a candidate). Look for Across DFX on a record.

Was this helpful?

Related

Still stuck?

The support chat knows the page you are on and your plan, or write to a person.