Multi-Asset Market Making And Correlated Inventory

A market maker quoting a single stock only needs to worry about that stock's own risk. A market maker quoting a hundred stocks in the same sector, or a handful of correlated ETFs, faces a different problem: each individual position might look small and within its own cap, while the combined position, because the instruments move together, behaves like one much larger directional bet. Managing each name in isolation misses this entirely.

Why correlation changes the picture

Two positions that are each half the size of a single-name limit look safe on paper if you check them one at a time. But if the two instruments have a correlation close to 1, say, two large-cap tech stocks that tend to move together, being long both is nearly the same, risk-wise, as being long one position at the combined size. A firm-wide or desk-wide risk limit that only checks each instrument against its own cap will let a trader build a position that's individually compliant everywhere but collectively far past what any single limit was meant to prevent.

Worked example: two correlated names

A market maker has position caps of 20,000 shares in each of Stock X and Stock Y, both in the same industry with a historical correlation of about 0.85. On a day where sector-wide buying pressure lifts both names, the desk ends up long 18,000 shares of X and 16,000 shares of Y, both comfortably under their individual caps. But because the two stocks move together, the combined position behaves like being long roughly $18{,}000 + 0.85 \times 16{,}000 \approx 31{,}600$ "X-equivalent" shares of correlated exposure, well beyond what a 20,000-share single-name cap was calibrated to allow for one stock's worth of risk. A sector-wide selloff would hit both legs at once, in the same direction, producing a loss far larger than either position's individual cap implies.

Drag the correlation slider in the explorer above toward 1 and watch the scatter tighten into a line, that's the visual version of what happens to combined risk: as correlation rises, two "separate" positions increasingly behave as one.

Managing it in practice

Multi-asset desks track a portfolio-level exposure measure, often a correlation-adjusted or beta-weighted aggregate position, in addition to per-name caps, and set a portfolio limit that catches exactly the scenario above even when every individual leg is within bounds. Hedging follows the same logic: rather than hedging each name separately, the desk can often hedge the shared, correlated component of risk once, cheaply, with a single sector or index instrument, while handling only the small residual name-specific risk position by position.