How multi-manager platforms allocate risk

From the outside, a multi-manager firm looks like a pool of money split between teams. From the inside, the more useful unit is risk. Two teams given the same amount of capital can take wildly different amounts of risk with it, depending on what they hold and how they hold it.

So platforms tend to allocate a risk budget rather than a dollar amount. A budget might be expressed as an expected range of daily gains and losses, as limits on exposure to a sector or a country, or as a combination of measures. The capital follows from the budget, not the other way around.

Each portfolio team works inside a mandate: what it may hold, how concentrated it may become, and how much it may lose before something changes. These limits are agreed before the team starts, which matters. A limit set in a calm week is a structure. A limit invented in a difficult week is a negotiation.

Loss limits are the most visible part of the design. When a team's losses reach a threshold, its budget is typically reduced and a review begins. This is not a verdict on the people involved. It is a way of making sure that one team's difficult stretch stays one team's difficult stretch.

A central risk team sees what no single team can: the sum of everyone. Two teams may each hold a modest position that, combined, is the largest exposure on the platform. Ten teams may each lean toward the same broad factor without knowing the others are doing it.

The center's job is to measure those overlaps every day, net them where they cancel, and limit them where they compound. It can ask a team to reduce an exposure that is fine in isolation but crowded in aggregate. In some designs the center also holds offsetting positions of its own to bring the firm's overall exposures back inside its limits.

The case for many independent teams rests on their results being unrelated to one another. In ordinary markets that can hold. In stressed markets, relationships between assets tend to shift, and strategies that looked independent can start moving together.

That is why careful platforms test their own diversification under stress scenarios rather than relying on how teams have behaved in quieter periods. The question is not how related the teams are on an average day. It is how related they become on the worst days.

The structure asks a lot of everyone. Portfolio teams accept limits that a standalone manager might not. The firm pays for a large central infrastructure in risk, technology, operations and compliance. Investors accept a structure whose terms are set out in detail in offering documents.

What the structure offers in return is discipline that does not depend on any one person's judgment on any one day. Whether that trade suits a particular investor is a question for that investor and their advisers, with the offering documents in hand.¹