A Deep Dive is an evidence packet, not a chat answer.
Asking ChatGPT can give you a fluent summary. Start with Board (`/board`) to inspect exact Hub Asset and point-in-time PRIMARY Basket identity before opening cited work. Research opens no ticker-only dossier or provider read from an unbound symbol.
Every number and quote stays attached to its source, date, and verification status. If a claim cannot be re-checked against the primary source, the dossier says what is missing instead of turning it into confident prose.
Board is the governed identity and evidence-discovery view; Deep Dive reads exact-bound public and owner-saved evidence; Vault (`/vault`) preserves saved evidence packets; and API (`/developers`) documents the current machine contract. Basket identity remains carried context, never a Research destination or local taxonomy.
The Board is the free learning surface. The Vault is where completed dossiers stay collected and refreshed. Every other page lives in the footer or behind a locked page's tabs or disclosures, and there is no fifth main-nav door. The AI layer remains labeled interpretation, never a recommendation or a signal exported to Decision Terminal.
Filing-groundedPrimary filings and transcripts are part of the source plan before prose is written.
Dated and re-checkableCitations travel with the packet so an old read can be aged against new evidence.
Machine-verifiedQuotes and numbers are re-fetched and matched; unmatched claims fail closed.
Method
Read the method, one chapter at a time.
11 chapters
01
How the Board is organized
Hub identity and Basket membership stay separate from Research evidence.
Every Board row preserves Hub instrument and issuer identity, PRIMARY Basket membership, and the exact point-in-time publication revision. Hub display order is context, never a Research rank.
Research adds only source-roster state and cited-content availability. It does not open a ticker-only dossier or turn evidence, grades, or prose into a score, vote, rank, recommendation, or Decision Terminal input.
02
Where the data comes from
Structured company context for the Board; cited primary sources for the Deep Dive.
Board identity and membership come from the exact current Hub Asset registry and paired-clock point-in-time Basket publication. Uncovered rows stay visibly pending; Research does not create a fallback peer set.
A Deep Dive widens that to 7 source classes: filings, earnings calls, company materials, independent research, web, social, and market data. Every finding keeps its source and date so it can be re-checked.
03
How a Deep Dive changes the read
Cited evidence changes the interpretation, not the objective feed.
A Deep Dive can only make a stronger claim when it surfaces dated, cited evidence that supports the interpretation. With no evidence, the read says what is missing rather than filling the blank.
The output is research, not investment advice: facts, citations, bull/bear tension, and caveats. It is not a recommendation and it is not exported as a ranking signal.
04
How we integrate with MarketSchema
Research reads the platform; it never holds pricing, tokens, or auth.
Objective measurements, quotes, and entitlements are read from the MarketSchema platform. Research is a read-only client of that data — it displays platform status but never owns it.
Pricing, token metering, checkout, and authentication live on the platform, not here. The current Research surface starts no fresh paid or model-backed work while exact Hub identity admission is unavailable.
05
Hub measurement boundary
Published measurement state stays separate from Research content.
Hub owns Basket relative strength and every numeric Basket measurement. Research preserves Hub PENDING, UNAVAILABLE, and WITHHELD states and never calculates a substitute.
Board controls filter exact Basket membership and Research evidence availability. They do not change membership, calculate a score, or reorder rows as an opinion.
06
What Board controls do
Filter governed context without creating a taxonomy.
Choose a Hub Basket or a Research evidence state to narrow the visible rows. The control consumes exact Asset IDs, PRIMARY versus CONTEXT role, and point-in-time publication identity.
There is no standalone Lens taxonomy. Changing a filter never creates a peer group, score, rank, forecast, recommendation, or trade input.
07
Board vs Deep Dive evidence
The Board orients quickly; the Deep Dive does the cited work.
The Board is built for fast evidence discovery: governed Asset/Basket context and typed coverage state. A dossier door stays withheld until its request can revalidate the exact binding.
Dossier evidence is slower and richer: filings, transcripts, company materials, independent research, and machine-verified citations. Historical saved evidence remains in Vault, not as a hidden Board number.
08
Freshness and aging
Saved research is dated, and it ages against new evidence.
Every Deep Dive is stamped with the date of the evidence behind it. Saved records keep their citations and a source-quality snapshot, so a months-old read is never mistaken for a current one.
Reopen a saved record and it is aged against fresh evidence: stale citations and elapsed catalysts are flagged before an old Deep Dive read is reused.
09
What we never do
No fabricated numbers, no advice, no hidden opinion scores.
When evidence is missing, the read says so rather than guessing. A fact you cannot trace to a source is a fact we do not pretend to know — honest blanks beat confident fiction.
Numbers stay measurements. Opinions stay labeled content. Nothing is nudged to make a name look better, and no Deep Dive or AI interpretation becomes a bot-consumed score.
10
Fresh compute boundary
Fresh work fails closed until exact Hub identity and point-in-time Basket provenance reach admission.
Fresh Research computeno fresh spend · typed unavailable
No provider or model call is enabled from the current Research surface.
Fresh dossier or briefing work remains closed until the request preserves an exact Hub instrument ID, issuer ID, and point-in-time Basket publication. Historical owner bytes remain untouched.
No unbound ticker, local Theme, Lens, score, rank, or peer set can open a compute path.
Research output, not investment advice.
11
Terms used in the terminal
Definitions stay tied to the current Board, Dossier, and Vault contract.
Objective MarketSchema measurement context that Research may read behind the scenes; it is not the retail Board order and never changes because of a Deep Dive.
The structured context snapshot the paid research layer compares a Deep Dive against; it stays in the research layer and never changes MarketSchema’s platform measurement.
