Institutional Flows separates the part of capital flow that can be traced to identified vehicles from the part that cannot. What can be named is named; everything else is grouped together as unattributed.

The unattributed remainder is the part most readers get wrong. Corporate treasuries that publish nothing, sovereign holdings, funds, miners, trading desks, very large holders and ordinary individuals all sit inside it, netted against one another, and it is not a measure of retail.

What it actually measures

The identified share says how much of the total activity the named vehicles account for, and deliberately not which way they went. Direction is kept on a separate line, which is what lets the share be compared across inflow and outflow alike.

A high identified share during inflow means the advance is being led by capital that can be pointed at. It is easier to verify than a broad move and easier to see stop.

The two sides working together is the healthier balance. Named and unnamed capital both buying describes a market with more than one source of demand.

The part that cannot be named is not the crowd

It is natural to treat the identified side as the professionals and the remainder as everyone else, because that maps onto a story people already believe. The mapping is simply wrong.

The remainder holds sovereign holdings, mining operations, desks moving size for clients and treasuries that have never filed anything, alongside individuals. Reading a rise in it as a crowd arriving discards most of what is actually in there.

Institutional Demand: removing supply is not the same as raising price

A squeeze means identified buying exceeds what the network issues over the same stretch. The difference has to come out of supply that already existed, which tightens what is available.

A squeeze is often reported as though the arithmetic settles the matter: less supply, higher price. The first half is true and the second does not follow from it.

Existing holders can supply the difference from what they already own, and they frequently have. The squeeze changes where the supply has to come from; it does not establish that there will be too little, so the reading is a condition and not a forecast.

What it does not tell you

Coverage decides everything the view can see. A vehicle that is not tracked lands in the remainder by necessity, and the boundary is about visibility rather than about size.

Attribution is only possible where a vehicle publishes. Earlier stretches of the record have less of it, so the identified share is not comparable across the whole history.

The two sides are netted before they are shown. Opposite activity inside either group cancels out, so a quiet reading can sit on top of a lot of movement that the line never reveals.

How to read it

Inst-Led Inflow. The named vehicles account for most of the activity while money is arriving.

Broad-Based Inflow. Named and unattributed capital are buying together, the healthier balance of the two.

Offsetting (Inst Buying). The named vehicles are absorbing what the rest of the market is releasing.

Diverging (Inst Selling). Rare and worth attention: named vehicles selling into buying from everyone else.

Institutional Flows shares the Capital Flows dashboard with Institutional Demand, Free Float and Cohort Decomposition.

Common questions

What is in the unattributed part?

Every buyer and seller the dashboard cannot put a name to. Unreported treasuries, sovereign holdings, funds, mining operations, desks, whales and individuals are all in there together, which is why calling it retail is wrong.

Why is the share unsigned?

Because it measures how much of the activity the named vehicles account for, not which way they went. Keeping direction separate is what makes the share comparable in both conditions.

Why does offsetting matter?

Because one identifiable set of buyers is taking everything the rest of the market lets go. That concentration has sat near turning points in both directions.

Can the identified share be compared across the whole record?

Not safely. Attribution depends on vehicles that publish, and the earlier record has far less of that, so the share reflects visibility as much as behaviour.

What does a quiet reading hide?

Potentially a great deal. Both sides are netted before they are drawn, so opposite activity within a group cancels and leaves a calm line over a busy market.

What sits on each side of this?

Buying that can be identified on one side, supply the network newly issues on the other, over the same stretch. What comes out is a balance rather than a level.

Is unidentified buying counted?

Not on the demand side, though all of the issuance is counted on the other. The balance therefore understates total demand and the gap is not measurable.