The Outcome Lifecycle: Predict, Underwrite, Deploy, Monitor, Resolve
Five stages, two functions, one closed loop. Why prediction and underwriting cannot be the same act, and why the loop breaks the moment resolved outcomes stop flowing back into the model.
Read →Calibrated Probability Is the Institutional Standard
What separates a measured probability from a confident guess, how calibration is actually tested, and what an institutional counterparty should demand before pricing capital against someone else’s number.
Read →Regulated Outcomes as an Asset Class
Litigation, regulatory enforcement, and insured claims are three professions built on the same structure. That structure, not the subject matter, is what makes a category underwritable.
Read →Why Duration Is the Underpriced Variable
A multiple hides the one variable that moves an annualized return more than almost any contested merits question. Why duration needs its own distribution, not a placeholder date.
Read →Settlement Distributions Versus Point Estimates
A single settlement figure is a guess wearing a measurement's clothes. What a labeled distribution actually claims, and why the difference changes how capital should be sized.
Read →The Two Audit Trails: Predict and Underwrite
A prediction record and an underwriting record answer different questions and have to exist as separate, timestamped artifacts, or no one can tell a bad model from a bad price after the fact.
Read →Calibration as a Governance Object
Calibration is a statistical property until an institution has to govern it. What ownership, versioning, and escalation actually look like once a model is running against live capital.
Read →What a Sealed Baseline Is For
Without a frozen reference point, 'we got better' and 'nothing material changed' are both unfalsifiable claims. What a sealed baseline is, and the three places it actually matters.
Read →Why Realized Outcomes Activate Claims Per Asset Class
Neither claiming realized-dollar prediction everywhere nor ruling it out permanently is correct. Why the claim activates one asset class at a time, on preregistered evidence.
Read →Concentration Risk in Legal Portfolios
A legal portfolio can look diversified by case type and jurisdiction while concentrating in a single judge, a single legal theory, or a single defendant. The four axes a standard credit framework misses.
Read →The Borrowing Base for Legal Receivables
Asset-based lending's borrowing base structure extends to legal receivables, but the collateral behaves nothing like an invoice. What eligibility, advance rate, and re-scoring have to look like.
Read →Medical Lien Behavior as a Data Asset
Medical liens are usually treated as an administrative line item. Examined at scale, lien behavior is a structured signal for injury severity, settlement timing, and net-to-claimant value.
Read →Judgment Portfolios as an Asset Class
A judgment is not a smaller version of the claim that preceded it. Why appellate and collection risk require their own pricing discipline, distinct from pre-resolution litigation finance.
Read →What Courts Can Learn From Outcome Data
Most of this market asks what capital can learn from courts. The reverse question, what courts can learn from their own outcome data, is underexplored and carefully bounded.
Read →Regulated Markets as the Unit of Analysis
Practice area is the wrong unit for organizing legal outcome intelligence. Why the regulated market a matter moves through, not its subject matter, determines what a model can actually learn.
Read →The Case Against AI Confidence Scores
A confidence score and a calibrated probability can look identical and mean nothing alike. Why institutional decisions should treat an unlabeled confidence score as a self-report, not a measurement.
Read →Disclosure Regimes and What They Change for Capital
Litigation funding disclosure requirements are usually filed under compliance. They belong in underwriting, because they change the negotiating dynamics and the structures a jurisdiction will actually support.
Read →Jurisdiction as a Pricing Input
Jurisdiction is usually a go/no-go screen at intake. Priced correctly, it is a continuous, granular input that shifts outcome and duration distributions more than most case-specific facts.
Read →The Role of the Human Decision in an Automated Pipeline
A rules-based, non-generative pipeline can carry a matter from raw data to a structured output. Where the human decision belongs in that chain, and why automating it away hides accountability rather than removing it.
Read →Why the Corpus Must Be Deduplicated Before It Is Counted
A raw record count is one of the easiest numbers to inflate by accident. Why every downstream claim, including a base rate's real sample size, inherits a corpus's duplication.
Read →ABS Structures and Law Firm Economics
Alternative business structure frameworks get discussed as a governance question. They are a capital-formation event, converting a law firm's financing menu from debt and receivables into a full stack that includes equity.
Read →What an Institutional Data Room Should Contain
A data room for a model-driven outcomes platform has to make the underlying data science diligenceable, not just the financials. Five sections most data rooms skip.
Read →Insurance and Litigation Finance Convergence
Insurance and litigation finance are converging, not around a shared product, but around a shared requirement for calibrated outcome and duration data neither industry historically had in-house.
Read →The Economics of a Second Deployment
A platform's real underwriting edge is invisible in its first deployment. Why construction cost, not marginal cost, dominates the economics of an early cohort, and what that means for new entrants.
Read →