Real-time holdings
Hypertable-backed positions table; latency well under 100ms for a typical advisor book.
Feature · Portfolio
Holdings, performance, and transactions on one screen.
TWR for performance. MWR for client-experienced returns. TimescaleDB hypertables for billions of position rows. Pricing service in Python (QuantLib). All locale-aware.
What it does
One context owns the position story.
The portfolio bounded context owns six aggregates: Account (open / closed / suspended), Instrument, Position, Transaction, Price, and the per-account PerformanceSnapshot. Positions live in a TimescaleDB hypertable partitioned by month and indexed on (tenant_id, account_id, asof) — billions of rows answer holding queries in tens of milliseconds.
Performance is computed by the pricing service, written in Python around QuantLib and numpy. Time-Weighted Return uses the Modified-Dietz approximation; Money-Weighted Return uses Newton-Raphson IRR. Both calculations cache to Redis with a tenant-scoped key, invalidated on the day’s transactions and end-of-day price updates.
Allocation analytics roll up by asset class, sector, geography, currency, and instrument type. The same data renders correctly for an English-speaking US advisor (USD, sector-by-GICS) and an Arabic-speaking Saudi advisor (SAR, sector-by-Tadawul). Concentration limits and house views are policy data; OPA evaluates them per query and per pre-trade check.
Reporting outputs through the document pipeline. Carbone renders the DOCX, Gotenberg converts to PDF, the formatter pack injects locale-correct numbers and dates. The PDF lives in MinIO with a SHA-256 content address; the audit chain references that address, not the bytes — so you can prove the statement existed at a point in time without copying it twice.
Capabilities
Eight things this context owns.
TWR performance
Modified-Dietz approximation; daily, MTD, QTD, YTD, since-inception. Periodicity selectable.
MWR / IRR
Newton-Raphson IRR for client-experienced returns; cash flow timing matters.
Allocation rollups
By asset class, sector, region, currency, instrument. House views applied per OPA policy.
Transactions ledger
Every settlement, dividend, fee, FX adjustment. Reconciled against the custodian feed daily.
Pricing service
Python + QuantLib; supports equities, bonds, ETFs, sukuk, FX forwards, and basic derivatives.
Statements & reports
Locale-aware via Carbone; sent through DocuSeal for client acknowledgement when required.
Custodian reconciliation
Daily Temporal workflow; cases opened automatically on persistent breaks.
TWR vs MWR
Which return are you reporting?
| Metric | What it answers | How we compute | When we use it |
|---|---|---|---|
| TWR | How well did the manager perform, ignoring the timing of client cash flows? | Modified-Dietz approximation; sub-period geometric linking | Manager scorecards; benchmark comparisons; advisor performance |
| MWR | What return did the client actually experience, including their own deposit and withdrawal timing? | Newton-Raphson IRR; iterates until residual < 1e-7 | Client statements; "what did I make?" answer; tax considerations |
| Cash-on-cash | Simple beginning-to-end return for the period | End MV − Begin MV − Net flows, ÷ Begin MV | Quick previews; not for client-facing reports |
Portfolio doesn't ship alone.
Feature
Trading
Orders flow into the portfolio context as fills; the position table updates as ExecutionReports settle.
FeatureDocuments
Statements, suitability summaries, performance reports — all generated through the locale-aware pipeline.
FeatureAudit
Every position write, transaction, and price update emits an audit event. Tamper-evident by design.
Portfolio walkthrough
See your AUM in Modir.
Bring a sample of accounts, transactions, and prices. We'll show you exactly how Modir renders them — in your locale, with your reporting standards.