A 3× return in
18 months is not the
same investment as
a 3× return in five years.

Every judge has a tempo. Some judges rule on motions to dismiss within 60 days. Others take 9 months. Some judges hold monthly scheduling conferences that keep cases moving. Others set an initial case management conference and do not intervene again until trial is imminent. These differences are not random — they are stable, measurable characteristics of individual judges that persist across cases and across years. The IAALS study confirmed this empirically: courts that were fast overall were fast at every stage, and courts that were slow overall were slow at every stage. The pattern is systemic, not episodic. The Judicial Pace engine profiles every judge using docket data from the federal CM/ECF system and state court electronic filing systems. For each judge, the engine calculates phase-by-phase processing times: median days from filing to ruling on a motion to dismiss, from MTD ruling to Rule 16 conference, from conference to discovery cutoff, from discovery cutoff to dispositive motion ruling, and from dispositive motion ruling to trial date. These calculations are segmented by case type — a judge who processes patent cases efficiently may handle employment cases slowly, and the duration model must reflect the judge's pace in the specific case type being funded. The engine also identifies temporal anomalies in a judge's recent docket: if a judge who historically rules on MTDs in 4 months has not ruled on the last three MTDs for 7+ months, something has changed — perhaps a senior status transition, a health issue, or a surge of complex cases. The anomaly is flagged and the duration projection is adjusted accordingly. When the investment committee reviews a funding opportunity, they see not just "estimated 24 months" but "assigned Judge Smith, who processes patent cases at a median of 22 months, with MTD rulings in 4.2 months versus the district average of 6.1 months, and no recent pace anomalies detected." That specificity transforms a guess into a calibrated projection.

A total case duration of "28 months" is an aggregate that conceals critical detail. Where in those 28 months does the risk concentrate? Is the uncertainty in the pleading phase (will the MTD be granted?), the discovery phase (how many custodians, how many documents?), the dispositive motion phase (how long until the MSJ ruling?), or the trial phase (will the court's calendar delay the trial date)? Each phase has its own duration distribution, its own risk factors, and its own sensitivity to external events. The Phase Decomposition engine models each phase independently. The pleading phase duration depends on whether the defendant files an MTD (probability assessed by the Merit engine), how long the assigned judge takes to rule on MTDs (from the Judicial Pace engine), and whether the ruling requires an amended complaint (adding 2-3 months). Fact discovery duration depends on the document universe size, the number of custodians, the complexity of the privilege review, the number and location of depositions, and whether foreign discovery (via Hague Convention letters rogatory) is required. Expert discovery duration depends on the number of expert reports, the complexity of the technical issues, and whether Daubert challenges are anticipated. Dispositive motion duration depends on the judge's historical MSJ ruling timeline and whether the motion requires a hearing. Pre-trial and trial duration depend on the court's calendar congestion, the estimated trial length, and the probability of a continuance. Each phase produces its own duration distribution (P25/P50/P75), and the phases are aggregated — accounting for dependencies between them — to produce the total case duration distribution. The result shows the investment committee not just the total projection but where the time goes and where the risk concentrates.

Discovery is the phase most vulnerable to duration expansion — and the phase where expansion is most predictable from case characteristics. A case with 500,000 documents from 8 custodians has a fundamentally different discovery timeline than a case with 15 million documents from 200 custodians across 5 countries. Yet most duration estimates treat discovery as a fixed phase rather than a variable one. The Discovery Complexity engine models the relationship between discovery scope and duration using regression models trained on thousands of completed cases. The primary inputs are document volume (the estimated number of documents in the discovery universe, including email, files, messaging platforms, and cloud storage), custodian count (each custodian represents a data collection, processing, and review stream that adds time), data source complexity (standard email and file servers versus mobile devices, encrypted messaging, foreign-language documents, and legacy systems), privilege review burden (the number of documents requiring privilege log entries, which scales non-linearly with document volume), and deposition logistics (the number of depositions, whether they require travel, foreign-language interpretation, or coordination with third parties). The model outputs a discovery phase duration distribution that accounts for these factors and their interactions. A case with 4.7 million documents and 47 custodians in a standard domestic commercial dispute has a different discovery duration than a case with the same document volume but 12 foreign custodians requiring Hague Convention procedures. The model captures these differences with empirical precision rather than qualitative judgment. The discovery phase is also where the most common source of delay events occurs: discovery disputes between the parties that require judicial intervention, extensions of discovery deadlines, and scope expansions driven by new information uncovered during the discovery process itself.

Dispositive motions are the binary gates of litigation — the moments where the case either survives to the next phase or ends entirely. From a duration modeling perspective, they create branching timelines: if the motion is denied, the case continues on one trajectory; if granted, the case ends (total loss for the funder); if granted in part, the case continues on a narrower trajectory with a modified duration profile. The Survival Gate engine models these branching points with their associated probabilities and timeline implications. For the motion to dismiss: the engine estimates when the motion will be filed (typically within 60 days of service), how long the assigned judge will take to rule (from the Judicial Pace engine), and the probability of each outcome (full denial, partial grant, full grant). Each outcome creates a different subsequent timeline. A full denial means the case proceeds to discovery on the original scope. A partial grant narrows the claims and may reduce discovery scope, shortening the subsequent phases. A full grant ends the case — but not always permanently: the claimant may file an amended complaint or appeal the dismissal, creating a new timeline branch. For summary judgment: the engine models MSJ timing relative to the discovery cutoff, the judge's historical MSJ ruling pace, and the probability of each outcome. Critically, cases that survive MSJ show a statistically significant increase in settlement activity — MSJ denial is a strong settlement catalyst because it signals to the defendant that the case will reach trial, changing the defendant's risk calculus. The engine incorporates this settlement acceleration effect: a case that survives MSJ has a projected total duration that is shorter on average than the pre-MSJ projection suggested, because the MSJ denial itself accelerates resolution.

The trial date is the most visible milestone in a case timeline — and one of the most unreliable. Courts set trial dates early in the case management process, but those dates are frequently continued due to calendar congestion, party requests, judicial reassignment, or events that require additional pre-trial preparation. A fund that models duration based on the initially scheduled trial date will consistently underestimate case length. The Calendar Congestion engine models trial date reliability by analyzing the assigned court's recent trial calendar: how many trials were scheduled in the last 12 months, how many actually proceeded on the scheduled date, how many were continued, and what was the average delay when continuances occurred. Courts with heavy criminal dockets tend to continue civil trials more frequently because criminal cases have speedy trial requirements that preempt civil cases. A court with 20 criminal trials scheduled in the next quarter may bump several civil trials to accommodate them, adding 3-6 months to the civil case timeline. The engine calculates a calendar congestion factor — a multiplier applied to the base trial date estimate. A factor of 1.0 means the court reliably holds trial dates. A factor of 1.4 means the court's civil trial dates are delayed by an average of 40% from the initially scheduled date. The continuance probability is modeled per judge: some judges grant continuances readily, others almost never. The combination of congestion factor and continuance probability produces a trial date distribution that captures the full range of when the trial will actually occur — not just when it is scheduled.

A favorable verdict is not the end of the investment. It is the beginning of the most dangerous duration risk: the appeal. A defendant who loses a $30 million verdict has every incentive to appeal — the cost of the appeal ($500K-$2M) is small relative to the potential savings if the verdict is reversed or reduced. And appeals take time: the median duration from notice of appeal to appellate decision varies by circuit, ranging from 10 months in the fastest circuits to 24 months in the slowest. During this period, the funder's capital remains deployed, the return is uncertain, and the IRR erodes with every passing month. The Appeal Probability engine models this risk across three dimensions. First, the probability that the losing defendant will appeal, based on the verdict amount (larger verdicts are more likely to be appealed), the legal issues presented (novel legal issues have higher appellate interest), and the defendant's financial capacity to fund an appeal. Second, the expected duration of the appeal, based on the circuit court's median processing time for civil appeals, the complexity of the issues on appeal, and whether the appeal involves interlocutory issues that may accelerate or delay proceedings. Third, the probability that the appeal changes the outcome: full reversal (the verdict is overturned), partial reversal (the verdict is reduced), remand (the case is sent back to the district court for further proceedings, potentially adding years), and affirmance (the verdict stands and the investment resolves). For funders evaluating appellate monetization opportunities — purchasing a share of a verdict currently on appeal at a discounted price — the appeal duration and outcome probabilities are the primary underwriting variables. A verdict with a 90% affirmance probability and a 10-month expected appeal duration is a fundamentally different investment from a verdict with a 60% affirmance probability and a 20-month expected duration, even if the nominal verdict amounts are identical.

Not all months in a case's timeline carry equal settlement probability. Settlement activity clusters around procedural milestones that change the parties' risk calculus: the completion of discovery (both sides now know the evidence), the ruling on summary judgment (the case is either proceeding to trial or it is not), the Daubert ruling (the expert testimony is either admitted or excluded), and the approach of the trial date itself (the cost and uncertainty of trial concentrate minds). The Settlement Timing engine models these acceleration windows for each case, based on the specific procedural milestones the case will encounter and the comparable-matter settlement patterns around those milestones. In a typical commercial dispute, settlement probability peaks in the window between the close of fact discovery and the ruling on summary judgment — months 10-14 in a 24-month case. During this window, 62% of comparable cases settle, because both sides have completed their factual investigation and can accurately assess their positions, but the cost of trial preparation (expert reports, pre-trial briefing, jury research) has not yet been incurred. A rational defendant who recognizes after discovery that the case is strong will settle now rather than spending $1-2 million on trial preparation and facing a jury verdict. The engine identifies the specific settlement windows for each case and models the probability that the case will resolve during each window. If the case settles in the early window (months 10-14), the total investment duration shrinks from 28 months to 12-14 months, dramatically improving the IRR. The duration distribution incorporates these settlement acceleration probabilities, showing the investment committee not just the trial timeline but the more likely resolution timeline — which, for 62% of comparable cases, is significantly shorter than the trial date would suggest.

This is where every other engine's output converges into the number that matters: the internal rate of return, adjusted for duration. The Duration-Adjusted Return engine takes the probability-weighted duration distribution (from Engines 01-07) and the probability-weighted outcome distribution (from the parent Capital platform's Return Structure engine) and calculates the IRR for every combination of duration and outcome. The output is not a single return number. It is a matrix showing the IRR at each duration percentile and each outcome scenario. At the P25 duration (16 months, reflecting early settlement in the acceleration window) with a base-case settlement of $24M on an $8M investment, the IRR is 78%. At the P50 duration (28 months, reflecting a trial-track resolution) with the same outcome, the IRR is 41%. At the P75 duration (38 months, reflecting trial plus a brief post-trial process) with the same outcome, the IRR is 27%. At the P95 duration (48 months, reflecting trial plus appeal), the IRR drops to 19%. Same case. Same 2.8× MOIC. Four radically different investment-quality assessments. The sensitivity analysis shows the investment committee which variables have the greatest impact on IRR: typically, the probability of early settlement (which compresses duration dramatically) and the probability of appeal (which extends duration substantially). A case where the settlement probability is high and the appeal risk is low will have a tight IRR distribution — the committee can be confident in the return. A case where the settlement probability is low and the appeal risk is high will have a wide IRR distribution — the committee faces genuine uncertainty about whether this is a 78% IRR investment or a 19% one. That uncertainty, quantified and visualized, is the engine's most valuable output.