FND003 · Network Intelligence

Beneficial Ownership Intelligence

FND003 Beneficial Ownership Intelligence is the foundation capability that traverses ownership and control edges backwards from a subject entity to the parties standing at the top of its chains, computing each one's effective stake and the confidence of the path that establishes it.

FND002 draws the edges. FND003 walks the ownership subset of them into chains and reports what stands at the end.

Risk indicators

Risk

• Deep chain whose multiplied stake falls below the twenty-five percent threshold
• Chain reaching exactly six hops, where traversal stopped rather than finished
• High path confidence averaged over one weak edge
• Circular ownership, leaving the structure with no terminal
• Ownership edge carrying no stated percentage
• Effective ownership computed over a chain with missing percentages
• Single owner sharing many target entities

Investigation actions

1. Read the ownership depth before the effective percentage
2. Treat a low stake at depth three as an artefact of multiplication, not dilution of control
3. Check the chain depth against the six hop limit before concluding no owner exists
4. Read the per-edge confidences rather than the path score for any chain you rely on
5. Establish the loop manually wherever circular ownership is indicated
6. Check whether missing percentages rather than genuine dilution produced a low stake
7. Look at the connected entities before treating a shared owner as a network

Detection logic

Traversal runs the graph in reverse, from the subject upward to parties with no incoming ownership edge, which are the chain terminals. Two governed constants bound it:

ownership_threshold 25.0 percent, the effective stake at which a terminal is reported as a UBO

max_depth 6, the maximum number of hops traversed from the subject

Effective ownership is multiplicative, and this is the single most important thing to know about the product. A chain of sixty percent through sixty percent through sixty percent yields 21.6 percent, which is below the twenty-five percent threshold. A party exercising control at every level of a three-layer structure is therefore not reported as a UBO at all. The arithmetic is correct and the picture of control is incomplete.

That divergence is why the layered ownership indicator fires on depth alone, independently of any percentage. The depth indicator exists precisely to catch what the threshold drops, and it is the reason a structure can be flagged while reporting no beneficial owner.

A maximum depth of six means a chain of seven is not traversed. An owner beyond six hops is not reported as absent — it was never looked for. Read a result carrying no identified UBO as no qualifying owner within six hops, never as no owner.

Path confidence is the arithmetic mean of the confidence scores on the edges in the path, rounded to four places, and the result level confidence is the mean of the path confidences. A mean dilutes a weak link: a five-edge chain carrying one edge at 0.30 and four at 0.95 averages 0.82 and clears the 0.7 low-confidence gate. The whole chain is then reported as well evidenced when one of its hops is not. Where a chain matters, read the edge confidences rather than the path score.

Four ownership risk indicators are emitted, sorted and deduplicated:

Layered ownership structure, where a path reaches depth three or more

Majority beneficial ownership, where effective ownership reaches fifty percent

Low confidence ownership path, where path confidence falls below 0.7

Circular ownership indicator, where the entity appears in its own chain

Circularity is reported as a boolean only. It tells you that an entity owns itself through some route, not which route.

A separate pass identifies owners meeting the twenty-five percent threshold against more than one target entity, reporting the connected count and an average confidence. This is the input to PAT002 Shared Beneficial Owner. FND003 establishes the overlap; PAT002 argues about whether it matters.

Curated from fc-03-network-intelligence-library/intelligence-products/specific/FND003/README.md