Skip to content
Relay
12 findings

Margaret Ellison, advisor (synthetic persona)Senior advisor, founder and executive practice, New YorkAbout $820M across 183 households; four relationships above $50M; 7 of those households are in this prototypeWho's who

Why this client

Thornbury: The family company raised a late-stage funding round

Why this client

Thornbury is number 4 of 11 on today's list: score 76 = materiality 76 × outside event weight 1.00 (firm default). Flagged on: Late-stage funding round recorded, day 5; Operating business stake, 77% of assets, illiquid; Legacy funded 0 of $4.0M. 3 passages from 1 document support it; 1 source disagrees, shown below.

Open Thornbury

Materiality comes from the agent that raised the item; the weight is yours to tune on Today's list. The ranking and the relevance scores are arithmetic, and a model never reorders either.

3

Cited passages, from 1 document

0

Related, not cited

1

Sources in conflict

0

Past review date

Why it was flagged

Outside event

  1. Outside eventLate-stage funding round recorded, day 5
  2. ClientThornbury household
  3. HoldingOperating business stake, 77% of assets, illiquid
  4. GoalLegacy funded 0 of $4.0M

Cited evidence, ranked

  • Concentration above a household's policy threshold can be reduced through staged sales under an existing trading plan, an exchange fund, or hedging, each with different tax and liquidity costs.

    Managing single-name concentration, passage p1prototype corpus, day 3

    Relevance 0.32="funding" +0.03"household" +0.03Cited by the opportunity record +0.25

  • A trading plan on its own does not restore a household to its policy threshold. It schedules sales, and the position stays above policy until they complete.

    Managing single-name concentration, passage p3prototype corpus, day 3

    Relevance 0.28="household" +0.03Cited by the opportunity record +0.25

  • Staged sales that realise short-term gains carry a higher tax cost than sales of lots held longer than one year.

    Managing single-name concentration, passage p2prototype corpus, day 3Cited, but shares no term with this opportunity

    Relevance 0.25=Cited by the opportunity record +0.25
How relevance is scored, and where a model would sit
The query is the opportunity as the record states it: its title, class, strategy, action and reason path, and nothing more. Each passage's score is the share of the query's weight it covers, where a rarer term across the corpus counts for more and a term the corpus has never seen counts as its rarest, plus 0.25 when the record cites the document, halved when the document is past its review date. The floor is 0.20. In production a model rewrites the query and reranks the candidates; the floor, the citation requirement, the exclusion of superseded documents and the refusal stay in code, so a verdict never depends on a model's prose.

Sources that disagree

Two current documents disagree: concentration plan suffices

Concentrated positions under a trading plan is past the review date. The newer document is more likely to be current, but nothing here has been withdrawn, so both stand until a desk resolves it.

no: a trading plan schedules sales and the breach stands until they complete

Managing single-name concentrationday 3CurrentNewer

yes: the plan is the mitigation and no further action is required

Concentrated positions under a trading planday -45Past review date

Relay shows both and applies neither. The advisor, or the desk that owns the older document, resolves it.

What the query matched

  • household2 passages
  • funding1 passage
  • assetsnone
  • businessnone
  • companynone
  • daynone
  • familynone
  • illiquidnone

What the team already knows

  • Call, 14 days ago: Discussed the company's late-stage funding round
  • Client service associate, today: K. Thornbury's CFO emailed the capital call notice this morning.

Full briefing, with what could not be established

Client picture · Documents