Account Aggregation

Account aggregation is the practice of connecting those so they can be seen together.

The reason it matters is not convenience, and the convenience framing is why it is usually undersold.

Why it is an analytical necessity rather than a tidiness feature

The first article of Pillar 31 establishes that a portfolio's risk is a property of the collection rather than of its parts — a holding's contribution depends entirely on how it moves against everything else held.

That has a direct consequence for anyone holding assets in more than one place: the individual account views are each incomplete in a way that cannot be fixed by looking at them more carefully. Concentration that is invisible within each account can be severe across them. Two accounts that each look diversified can hold the same exposure twice, and neither statement will say so because neither statement can see the other.

The strongest case for aggregation, and it is not about saving time. Every measure in Pillar 31 — correlation, the diversification floor, a position's share of total variance, geographic and currency exposure — is undefined for a portfolio you can only see in pieces. A position that is 8% of one account and 8% of another is not an 8% position; it may be 15% of the whole, and the position-sizing arithmetic shows that its share of risk rises faster than its share of the money. Aggregation is therefore the precondition for the analysis rather than a presentation of it. This is a statement about arithmetic, not a reason anyone should use any particular product.

How connections actually work

Four mechanisms are in general use across the industry, and they differ enormously in what they require and what they risk. They are described here as categories, not as a description of any particular platform's implementation.

Direct data-sharing interfaces. The provider exposes a permissioned interface, the user authorises access at the provider, and the aggregator receives a revocable token. Credentials are never shared with the aggregator, and access can be withdrawn by the user at the provider's end. This is the strongest arrangement and its availability depends on the provider.

Credential-based access. The user supplies their login details, which are stored and replayed to retrieve data. The user's credentials leave their control, the access is not scoped, and revocation generally means changing the password. This mechanism predates the alternatives and persists where providers offer nothing better.

File import. Statements or transaction exports uploaded by the user. No standing access exists at all, at the cost of being manual and periodic.

Manual entry. Complete control, no connection, and the maintenance burden is the reader's.

What can go wrong, stated fully

Security. Any standing connection is an additional place where something about the account exists. The material question is whether credentials or a revocable token are involved, since the two fail very differently.

Scope. Read access and transaction access are different permissions, and a reader should know which one they have granted to anything.

Breakage. Connections fail when providers change their systems, and a connection that silently stops updating is worse than no connection, because the view still looks complete.

Data quality. Cost basis is frequently wrong or absent on aggregated holdings, corporate actions may be handled inconsistently, and transactions can be duplicated or missed. Aggregated data is a good enough basis for seeing exposure and a poor basis for anything requiring exact records — tax reporting above all, which remains the province of the providers' own documents and is parked to Annex A.

What MarketClue does, and what it deliberately does not

MarketClue's aggregation is read-only by design. It exists to let a reader see holdings together so that portfolio-level measures can be computed on the whole rather than on fragments.

What it does not do is the more informative half. MarketClue cannot place, modify or cancel an order on any connected account, cannot move money, and cannot change any setting at any provider. No connection MarketClue holds carries transaction capability, because the platform has no function that would require it.

MarketClue is not a broker — it holds no assets and executes nothing, so a connection to it is a connection to a viewer rather than to an operator.

And connected data is not used to sell anything. MarketClue does not sell or share holdings data, does not use it to target advertising, does not carry advertising, and receives no payment from any provider for a connection or a placement. The platform is paid by its subscribers, which is the whole of its revenue model and is stated here because the first article in this pillar argues that a service is shaped by what pays for it — an argument that applies to the publisher as much as to anyone described.