Capital markets are good at pricing two properties of a position and consistently bad at pricing a third. Size, the amount of capital at stake, gets careful attention because it bounds the loss. Probability, the chance of a favorable resolution, gets increasing attention as outcome modeling matures. Duration, the time a position stays deployed before it resolves, is treated as background noise, folded into a multiple as if it were a fixed constant rather than a variable with its own distribution and its own error. This is backwards. Within the realistic range of case durations that any real portfolio holds, a shift in expected months to resolution moves an annualized return further than most of the merits questions that consume the bulk of underwriting attention. A position priced on the strength of its facts and mispriced on the strength of its clock will still underperform, and the underperformance will look, from the outside, like a merits problem it never was.
The arithmetic is not subtle once it is stated plainly. A multiple, the ratio of capital returned to capital deployed, says nothing about the rate at which that return accrued. Two positions can share an identical multiple and represent entirely different economics once time enters the calculation, because an eighteen-month resolution and a four-year resolution compress or stretch the same multiple into annualized returns that can differ by a wide margin. A capital allocator who prices from the multiple alone is pricing a number that hides the one variable most responsible for whether the position actually clears the cost of capital it was funded against.
Duration has been underpriced for a structural reason, not a careless one. Legal-asset finance inherited its pricing habits from a market that priced settlement value and cost of capital carefully and treated time to resolution as a fixed planning assumption, a single expected date rather than a distribution with a visible tail. A single date is easy to build a model around and easy to explain to a committee. It is also wrong in a specific, predictable direction: it understates the probability of the long tail, the small but real share of matters that take multiples of the median duration, and that tail is exactly where capital gets trapped longest and returns compress hardest.
A distribution-based duration model replaces the single date with a band and a stated confidence, built from a corpus of matters that actually reached resolution rather than from an analyst's intuition about how long litigation usually takes. The band has a shape, typically right-skewed, with a median duration well short of the mean because a minority of matters extend the distribution's tail considerably further than the bulk of matters cluster around. Reporting the median alone, the way informal duration estimates typically do, systematically understates the capital's true expected holding period, because it discards exactly the tail that drives the difference between an acceptable outcome and an actual loss.
Duration is not a static estimate made once at underwriting and left alone. A matter's procedural posture changes the distribution as it unfolds, and each change is a signal that should update the estimate rather than wait for it to be confirmed at resolution. A judge assignment changes the venue-level base rate for time to disposition. A motion ruling changes the range of remaining procedural steps. A counterparty's litigation conduct, delay tactics, aggressive motion practice, changes the shape of the tail directly. A duration model that is not re-estimated against these signals is not a model of duration; it is a static number wearing a model's vocabulary.
The institutional consequence of treating duration as background noise is a portfolio that looks diversified on outcome and is not diversified on time. A book can spread its capital across dozens of matters, different case types, different jurisdictions, different counterparties, and still concentrate its risk if every position in the book was underwritten to the same expected duration and that expectation turns out wrong in the same direction for correlated reasons, a court system slowdown, a change in a jurisdiction's procedural rules, a shift in how a category of defendant litigates. Correlation in duration is a distinct risk from correlation in outcome, and a portfolio construction process that only checks the second has not actually checked concentration.
Pricing duration correctly changes what a capital allocator can promise a limited partner or a balance sheet counterparty. A fund that reports expected duration as a single number is reporting a planning assumption dressed as a forecast, and its liquidity planning inherits that fiction. A fund that reports a distribution, with a stated tail probability and a re-estimation cadence disclosed alongside it, is reporting something an institutional investor can actually plan a capital call schedule against, because the investor knows not just the expected case but the range of cases the fund is actually exposed to.
A fund with a defined life adds a specific version of this problem that a single-position view does not surface. A fund with a fixed remaining term that underwrites a position whose duration distribution assigns real, disclosed probability to resolution beyond that term has taken on tail risk the fund's own structure cannot absorb, regardless of how favorable the position's outcome probability looks in isolation. This is not a hypothetical edge case; it is a standard feature of duration distributions with a meaningful right tail, and a fund that underwrites against expected duration alone, without checking the tail against its own remaining term, can find itself holding positions it structurally cannot wait out, forced into a discounted secondary sale or an extension negotiation that a duration-aware underwriting process would have flagged at the point of commitment rather than discovered near the fund's wind-down.
There is also a comparative dimension worth stating plainly: two institutions can hold portfolios with identical outcome-probability profiles and represent very different risk once duration is examined, because the institution whose positions cluster near the short end of their respective distributions can recycle capital into new positions faster than the institution whose positions sit further out on the tail, even if both institutions' matters resolve favorably at the same rate. Capital velocity is a direct function of duration, not of outcome probability, and an institution that reports only its win rate while leaving its duration profile undisclosed is withholding the half of its performance that determines how often that win rate actually compounds.
The institutions that treat duration with the rigor legal-asset finance has historically reserved for outcome probability will price risk correctly more often than the institutions that do not, not because their models are more sophisticated in every dimension, but because they are not silently mispricing the one variable that moves an annualized return more than almost anything else in the position. Duration is not the variable this market understands least. It is the variable this market has spent the least effort trying to understand, which is a different problem, and a solvable one.