Portfolio Construction Across Case Types and Jurisdictions

Case count is not diversification

A portfolio of two hundred cases filed in the same three counties, before a handful of the same judges, against a concentrated set of insurance carriers, is not diversified in any sense that reduces tail risk — it is two hundred draws from a small number of correlated risk factors. Correlation in litigation portfolios does not run primarily through "case type" the way an equity portfolio manager thinks about sector correlation. It runs through the judge assigned (docket-level rulings on class certification, summary judgment standards, and scheduling apply to every case before that judge simultaneously), through the venue (a single adverse appellate ruling on a legal standard affects every pending case relying on that standard in the circuit), through the defendant or defendant class (a single insurer's change in claims-handling posture affects every open claim against it at once), and through the underlying causation or liability theory (a single Daubert ruling on a causation methodology can affect an entire mass tort docket in one motion).

Underwriting each case independently and summing exposures produces a portfolio that looks diversified on a spreadsheet and behaves like a concentrated bet on four or five real risk factors when a common shock hits.

The axes that actually matter

A concentration framework built for litigation portfolios needs to track exposure along judge, venue, defendant or defendant class, and causation theory as first-class dimensions — alongside the case-type and jurisdiction dimensions most funds already track. A single case can carry meaningful correlated exposure on more than one of these axes simultaneously: the same MDL judge overseeing the same causation theory against the same manufacturer defendant is one risk factor wearing the appearance of many separate case files. Concentration limits set on case-type alone will not catch this; concentration limits set on the judge-venue-defendant-theory quadruple will.

Jurisdiction selection interacts with this directly. A fund that files aggressively in a small number of plaintiff-favorable venues to optimize expected value on each individual case is, at the portfolio level, concentrating docket-level and judge-level risk in exchange for that expected-value edge — a trade that needs to be sized and priced explicitly, not treated as a free optimization.

Building the correlation matrix instead of assuming independence

Practically, this means underwriting systems need to store structured case metadata — judge, court, defendant entity (resolved to parent company, not just the named defendant), and causation theory tag — as queryable fields, not free text buried in case notes, so that concentration can be computed across the live book at any point rather than reconstructed manually before an LP report. It also means stress-testing the portfolio against specific correlated scenarios rather than only against aggregate loss-rate assumptions: what happens to the book if this specific judge is reversed on appeal in a way that affects docket-wide scheduling, or if this specific causation theory is excluded under Daubert in the lead bellwether. Those scenario tests reveal concentration that a portfolio-wide expected-value calculation will not.

What to do once concentration is visible

Finding concentration is only useful if it changes behavior. In practice, funds that take concentration reporting seriously respond in one of three ways once a judge, venue, or defendant cluster crosses a defined threshold: they pause new originations into that specific cluster until existing exposure resolves or is repriced; they seek co-investment or syndication partners to lay off part of the concentrated exposure rather than carrying it alone; or they explicitly reprice new deals in that cluster at a return premium that compensates for the correlation the fund is choosing to keep on its book.

What does not work is treating the concentration report as a compliance artifact reviewed quarterly and acted on rarely. The value of the reporting is in the response cadence matching the rate at which new deal flow can add to an existing cluster — for an active origination desk, that rate can be faster than a quarterly review cycle catches, which means the report needs an owner who can act on it between formal reviews, not only during them.

The same discipline applies when a fund is evaluating a new co-investment or syndication opportunity offered by another manager: reviewing the counterparty's own concentration profile, not just the specific position being offered, is the only way to know whether accepting the position adds a genuinely diversifying exposure or simply layers the fund's book on top of a concentration the counterparty is trying to lay off for exactly that reason. A position that looks attractive in isolation can be exactly the exposure a knowledgeable counterparty is trying to exit.

What to ask for from an intelligence provider

• 01Portfolio-level concentration reporting on judge, venue, defendant (resolved to parent entity), and causation theory — not case-type alone.
• 02Scenario stress tests built around specific correlated shocks (an adverse appellate ruling, a Daubert exclusion, a carrier posture change) rather than only aggregate loss-rate sensitivity.
• 03Structured, queryable case metadata rather than free-text case notes, so concentration can be recomputed as the book changes.
• 04An explicit statement of how jurisdiction-selection strategy trades case-level expected value against portfolio-level concentration.
• 05A pre-trade check for any new position against the fund's existing judge, venue, and defendant concentration, run before commitment rather than discovered afterward in quarterly reporting.