Garbage in, confident garbage out

The most dangerous output in wealth software is a clean-looking recommendation built on dirty data. A rebalance computed against stale prices, a tax-loss harvest planned without real tax lots, a performance chart drawn from incomplete history — each looks authoritative and each can be wrong. Connected accounts and data quality are the unglamorous foundation that decides whether everything above them can be trusted.

This is worth treating as a first-class, user-facing concern rather than plumbing hidden in the basement. Before you accept any portfolio, tax, planning, or reporting output, you should be able to see whether the data underneath it is current and resolved.

The connection path, step by step

Bringing an account in is a sequence, and each step changes what you can trust:

• Select a provider — a brokerage, custodian, bank, direct API, or managed-account source.
• Authorize access — through OAuth, a hosted widget, or a direct data contract.
• Import accounts — the account is classified by type and linked to the right household, client, entity, or mandate.
• Sync holdings — raw positions arrive with tickers, identifiers, quantities, prices, and provider fields.
• Resolve securities — holdings are mapped to a security master by FIGI, CUSIP, ISIN, ticker, or exchange, with a fallback when no canonical match exists.
• Enrich positions — prices, portfolio positions, and tax-lot bridges are updated where the source data supports them.
• Precompute performance — daily returns and cached views are built from snapshots and cash-flow history.
• Review freshness — you see whether downstream analytics are current enough to act on.

The order matters because trust accumulates along it. A position that synced but did not resolve is not yet something you should rebalance around.

Security resolution and the synthetic-security problem

A subtle but critical step is mapping raw holdings to real securities. Most resolve cleanly by identifier. But when no canonical match is found, the system must create a placeholder — a synthetic security — so the position is not simply dropped. That placeholder is a flag, not a fact: before you rely on analytics for a synthetic security, confirm its ticker, identifier, asset type, and exchange.

Identifier priority (FIGI, CUSIP, ISIN, ticker, exchange) exists precisely because tickers alone are ambiguous across markets. Getting identity right is what keeps two different instruments from being silently treated as one.

Data-quality states you should know

Healthy data infrastructure is honest about its own gaps. A few states every reviewer should recognize:

• Stale price — the latest trusted price is older than expected; treat performance, risk, and rebalance output as provisional.
• Snapshot missing — daily holdings history is incomplete, so performance precompute may be unavailable.
• Tax lot missing — lot detail wasn't supplied; harvesting and after-tax decisions need review before action.
• Synthetic security — a placeholder was created; verify identity first.
• Cache cold — precomputed views are absent or expired; the app may compute live or show an unavailable state.

These are not error messages to dismiss. They are the difference between provisional and trustworthy.

The review rule: warnings must travel

Here is the principle that ties it together: if a connected-account warning exists, it must follow the data into every downstream surface that depends on it. A stale-price warning belongs not just on the accounts page but on the portfolio view, the risk report, the rebalance proposal, and any action packet built from that data.

A clean recommendation with no data lineage is not acceptable. The point of surfacing data quality is to prevent a downstream screen from presenting confident output while quietly resting on a known gap. The warning is part of the answer.

Why each user class needs this

Different users lean on the same foundation for different reasons. A self-directed investor needs it as a basic trust check before acting. An advisor needs it as a client-review preflight — you do not want to present a portfolio review built on a half-synced account. A fund or institutional manager needs it as lineage for holdings, benchmark, mandate, and reporting decisions that have to withstand scrutiny.

The takeaway

Connected accounts and data quality are where portfolio truth is established or quietly compromised. A disciplined pipeline — sync, resolve, enrich, precompute, and check freshness — plus honest data-quality states and a rule that warnings travel with the data, is what lets everything above it be trusted. The most sophisticated analytics in the world are only as good as the holdings they run on, which is exactly why this layer deserves to be visible rather than hidden.