Devancore Inc.
Devancore Post-Trade Glossary
Glossary
Natural Language Portfolio Questions
Natural language portfolio questions let institutional users ask about positions, cash, exposure, performance, restrictions, and breaks while the answer remains grounded in controlled portfolio records.
Document source: https://devancore.com/glossary/portfolio-questions-natural-language/
Devancore Post-Trade Glossary
Natural Language Portfolio Questions
Natural language portfolio questions let institutional users ask about positions, cash, exposure, performance, restrictions, and breaks while the answer remains grounded in controlled portfolio records.
Definition
Natural language portfolio questions are controlled queries over institutional portfolio records. A user can ask about positions, cash, exposure, performance, restrictions, unsettled trades, fees, tax lots, corporate actions, or reconciliation breaks without writing SQL, building a spreadsheet, or waiting for a custom report.
The control problem is precision. A question that sounds simple can depend on different books of record. "How much cash do we have?" may mean settled cash at the custodian, projected cash after pending trades, accounting cash in ABOR, or available cash after restrictions and collateral. "What is our Apple exposure?" may mean issuer exposure, equity position, option delta, sector weight, currency exposure, or mandate concentration.
Portfolio question to source record
Portfolio question to source record
A useful answer starts by identifying the book, scope, time, and calculation basis.
| Question type | Record needed | Control question |
|---|---|---|
| Position | IBOR position, account, fund, security, quantity, location, trade date, and settlement date | Which book and date define the holding? |
| Cash | Settled cash, projected cash, ledger cash, currency, bank or custodian source, and pending movements | Is the answer using available, projected, or accounting cash? |
| Exposure | Position, price, FX, issuer, country, sector, strategy, derivative mapping, and risk factor | Which taxonomy and valuation source created the exposure? |
| Performance | PBOR return, benchmark, weights, prices, FX, cash flows, income, fees, and attribution method | Can the return explanation be reconstructed? |
| Restriction | Mandate rule, restricted list, account eligibility, concentration limit, override, and approval state | Is the question testing policy or only describing data? |
| Break | Reconciliation item, source records, owner, age, reason code, materiality, and evidence bundle | Is the difference open, explained, escalated, or closed? |
A natural-language answer should start by resolving four things: portfolio scope, book basis, time basis, and calculation basis. Scope determines which funds, sleeves, accounts, entities, or strategies the user is allowed to query. Book basis determines whether the answer comes from IBOR, ABOR, PBOR, custodian records, risk data, or a reconciled view across them. Time basis determines whether the answer is current, end-of-day, as-of a historical date, or as-at what the firm knew at a specific timestamp. Calculation basis determines how prices, FX rates, benchmarks, accrued income, pending trades, fees, and overrides are handled.
The distinction between record interrogation and action is important. A portfolio question asks what the controlled records show. It may reveal a break, restriction, liquidity issue, stale price, or exposure concentration. That does not make the answer an approved trade, accounting entry, cash movement, or exception closure. If the user wants to act on the answer, the workflow should create a separate draft, validation, review, and approval path.
Good portfolio answers include citations. A response such as "cash is lower because of unsettled purchases" is not enough for institutional operations. The answer should identify the accounts, currencies, trades, cash movements, settlement dates, source systems, timestamps, and unresolved breaks that support the statement. Without that evidence, natural language becomes another reporting surface that teams must manually verify.
Portfolio questions - record dependency map
Devancore · message matrix
| Rail | Message | Purpose | Record |
|---|---|---|---|
| Holdings | what do we own? | position answer | IBOR position, account, security, quantity, custodian, trade date, and settlement date |
| Cash | what can we use? | liquidity answer | settled cash, projected cash, ledger cash, pending movements, currency, and source timestamp |
| Exposure | where is the risk? | risk answer | issuer, sector, country, currency, derivative mapping, price, FX, and risk-factor data |
| Return | why did it move? | performance answer | PBOR return, benchmark, weights, cash flows, income, fees, attribution method, and exceptions |
| Breaks | what is unresolved? | operations answer | reconciliation item, source rows, owner, aging, reason code, materiality, and evidence |
How it works
Natural language portfolio questions work by turning a plain-language request into a governed data query. The workflow scopes the user, interprets the question, retrieves permitted records, calculates the answer, returns source-grounded text, and records the question-answer pair.
Natural-language portfolio query controls
Natural-language portfolio query controls
The workflow should convert a question into a deterministic answer request.
| Step | What the system resolves | Evidence retained |
|---|---|---|
| Scope | User, role, desk, fund, account, portfolio hierarchy, permitted fields, and data domain | Entitlement result and denied fields |
| Interpret | Intent, book basis, metric, date range, currency, aggregation, filters, and ambiguous terms | Parsed query, missing fields, and clarification state |
| Retrieve | IBOR, ABOR, PBOR, cash, position, price, benchmark, restriction, settlement, and break records | Source system, record IDs, version, and as-of timestamp |
| Calculate | Exposure, cash view, P&L, return, aging, limit result, or reconciliation status | Formula, input set, currency, price source, FX source, and calculation timestamp |
| Answer | Plain-language explanation with cited records and warnings | Displayed answer, cited IDs, stale-data flags, and user session |
| Route | Follow-up query, export, task draft, exception note, or action request | Boundary between answer, draft, approval, and downstream workflow |
Scoping runs before retrieval. A portfolio manager, operations analyst, controller, compliance user, and executive may have different portfolio access, field access, and permitted actions. The system should not retrieve records the user is not entitled to see and then rely on the model to hide them in the final answer.
Interpretation converts human phrasing into a structured answer request. The system should resolve portfolio name, account, security identifier, issuer, currency, date, aggregation, metric, and book basis. If the wording is ambiguous, it should ask for clarification or show the assumption. It should not silently choose between IBOR and ABOR, settled and projected cash, trade-date and settlement-date positions, or current and historical knowledge.
Retrieval gathers the source records needed for the answer. Depending on the question, the workflow may pull positions, trades, cash balances, prices, FX rates, benchmarks, classifications, restrictions, tax lots, settlement status, corporate actions, and open breaks. Each retrieved data set should carry source, timestamp, version where available, and record identifiers.
Calculation should be deterministic. AI can explain the result, but the math should come from defined calculation logic: sum positions, convert currency, aggregate exposure, apply benchmark weights, calculate return, age breaks, or compare cash views. The answer should show the as-of timestamp and the calculation basis so users know whether the number reflects current operating state, prior close, or corrected historical knowledge.
Answer generation translates the result into usable language. The answer should include citations, warnings, unresolved data-quality issues, stale-source flags, and drill-through records. If a citation references a trade, account, CUSIP, cash movement, or break ID, the system should validate that the referenced record exists before showing it.
Routing handles follow-up. The user may ask another question, export evidence, create a task, draft an exception note, or request an action. Those follow-ups should preserve the original question, retrieved context, answer, citations, and timestamp. Any order, correction, override, or close decision should move into its own approval workflow.
Portfolio answers - trust boundary
Devancore Glossary · devancore.com
In Devancore™
Devancore - portfolio question control loop
Devancore Glossary · devancore.com
Devancore supports natural language portfolio questions as a governed query layer over operating records. The platform can help users ask about positions, cash, exposure, performance, restrictions, settlement state, and reconciliation breaks while keeping the answer connected to source records.
Devancore should not be framed as an investment adviser, portfolio manager, broker, custodian, accounting authority, performance standard-setter, or autonomous trading engine. Its role is to help firms expose controlled portfolio records through permissioned, cited, and auditable workflows.
In a Devancore-style workflow, the user asks a question. The system checks entitlement, resolves portfolio scope, selects the relevant book, retrieves permitted records, applies deterministic calculations, generates an answer with citations, and stores the full question-answer record. The answer can then feed a task draft, exception workflow, reconciliation review, compliance check, or report without losing lineage.
This page complements the conversational finance cluster by focusing on the portfolio inquiry surface. Conversational finance describes the broader interface. Conversational books and records explains governed access to regulated records. Natural language portfolio questions explains how investment teams, operations teams, controllers, and risk users ask practical questions about portfolio state and receive answers that can be checked.
The useful product boundary is simple: answer with evidence, then route action through controls. A natural-language answer may reveal that cash is short, a position is stale, a restriction is breached, a performance number moved, or a break is aging. The platform should help the user understand that state, cite the records behind it, and move the next step into a controlled workflow.
Related terms
- Investment Book of Record
https://devancore.com/glossary/investment-book-of-record/
The IBOR — a real-time position record used by investment managers, capturing unsettled trades, accruals, and corporate actions ahead of custodian confirmation and ABOR settlement.
- Accounting Book of Record
https://devancore.com/glossary/accounting-book-of-record/
The ABOR: custodian-confirmed settled positions used as the authoritative basis for NAV calculation, financial statements, and regulatory reporting.
- Performance Book of Record
https://devancore.com/glossary/performance-book-of-record-pbor/
The PBOR — a position record that extends the IBOR with return attribution, risk analytics, and benchmark data, providing the authoritative basis for investment performance measurement and client reporting.
- Position Management Securities
https://devancore.com/glossary/position-management-securities/
The real-time tracking of a firm's securities holdings across all accounts and custodians, updated as trades execute, settle, and corporate actions are applied.
- Cash Reconciliation Software
https://devancore.com/glossary/cash-reconciliation-software/
Software that matches a broker-dealer's internal cash ledger against bank statements and clearing utility records in real time, surfacing breaks for resolution before they create reserve formula errors, missed sweeps, or Rule 15c3-3 violations.
- Position Reconciliation Software
https://devancore.com/glossary/position-reconciliation-software/
Software that automates daily comparison of internal position records against custodian statements, prime broker reports, and on-chain ledger state, surfacing breaks before they affect Rule 15c3-3 determinations, NAV, or securities count obligations.
- Portfolio Accounting Software
https://devancore.com/glossary/portfolio-accounting-software/
Software that converts portfolio transactions, positions, cash, income, accruals, prices, tax lots, and realized gains into controlled accounting and reporting views.
- Portfolio Accounting System
https://devancore.com/glossary/portfolio-accounting-system/
A portfolio accounting system is the controlled architecture that maintains transactions, positions, cash, accruals, valuation inputs, tax lots, reconciliation, and audit evidence for portfolio accounting workflows.
- Portfolio Risk Controls
https://devancore.com/glossary/portfolio-risk-controls/
Portfolio risk controls turn exposures, limits, liquidity, counterparty, collateral, and settlement signals into governed post-trade checks, exceptions, approvals, and evidence.
- Performance Attribution Record
https://devancore.com/glossary/performance-attribution-record/
A performance attribution record is the controlled operating record that explains portfolio return using positions, weights, prices, cash flows, benchmarks, classifications, FX, corporate actions, and evidence.
- Trade Reconciliation
https://devancore.com/glossary/trade-reconciliation/
The systematic comparison of internal trade and position records against external sources to identify breaks and resolve them before they become settlement failures.
- Trade Break Management
https://devancore.com/glossary/trade-break-management/
The exception workflow for identifying, classifying, and resolving post-trade discrepancies before they breach the T+1 affirmation cutoff or trigger CSDR cash penalties.
- Maker-Checker Workflow
https://devancore.com/glossary/maker-checker-workflow/
A two-person segregation of duties control requiring that any action entered by one operator must be reviewed and approved by a second before it takes effect.
- Broker-Dealer Audit Trail
https://devancore.com/glossary/broker-dealer-audit-trail/
The immutable, chronologically linked record of every trade lifecycle event — from order receipt through settlement — maintained to satisfy SEC Rules 17a-3 and 17a-4, FINRA clock synchronization requirements, and CAT reporting obligations.
- Segregation of Duties (SoD)
https://devancore.com/glossary/segregation-of-duties-financial-software/
Segregation of duties (SoD) is the internal control principle that no single operator can book, approve, and settle a transaction — enforced through conflict matrices, maker-checker workflows, and access certification reviews to satisfy SOX Section 404.
- Conversational Finance
https://devancore.com/glossary/conversational-finance/
Conversational finance is a controlled natural-language interface for financial records, workflow intent, approvals, and evidence across trading, post-trade, compliance, and reporting.
- Conversational Books and Records
https://devancore.com/glossary/conversational-books-and-records/
Conversational books and records are permissioned natural-language workflows for querying regulated records while preserving source citations, retention, supervision, and audit evidence.
- Natural Language Trading
https://devancore.com/glossary/natural-language-trading/
Natural language trading converts a user's plain-language trade intent into a structured order draft, validation record, human review step, routing instruction, and audit trail.
- AI Audit Trail Financial Services
https://devancore.com/glossary/ai-audit-trail-financial-services/
An AI audit trail in financial services records the prompt, permissions, retrieved context, model output, proposed action, human decision, downstream event, and final outcome for each AI-assisted workflow.
Contact
Request a briefing or platform walkthrough with the Devancore team.
Request access https://devancore.com/access/