Worked Examples

These are methodology walkthroughs, not blind predictions. Both cases below are ones we already knew the outcome of — they're in fact two of the real cases in our own citation database — so this is not a claim that the calculator "called it" in advance. What it shows is how the documented probability-and-damages framework reasons through real facts, including where a court's actual ruling landed inside, at the edge of, or beyond the range the methodology would produce.

A genuine, unbiased track record requires predicting a case's outcome before it's decided, on the record, and checking back later. That's exactly what our prediction registry does — timestamped predictions on real, currently pending matters, checked against the actual outcome once each case resolves.

The facts

A landlord sought $520,361.29 in rent it said the tenant, a gym operator, owed as of October 2021. The tenant raised force majeure, impossibility, impracticability, and frustration-of-purpose defenses tied to pandemic-era closures — arguing it shouldn't have to pay because it couldn't operate the gym as intended.

Applying the rule

Our documented unpaid_rent rule starts at 90–97% probability for an essentially undisputed debt, dropping to 55–75% once a tenant raises a substantive dispute (for example, an abatement claim tied to the landlord's own breach). A defense was raised here, so a mechanical read says: apply the lower, disputed-debt range.

But applying the rule well means judging the strength of that defense, not just whether one was raised. By the time this case was litigated, force-majeure and impossibility arguments against paying rent — as opposed to operating a business — had already been broadly rejected in reported COVID-era commercial lease decisions. A defense procedurally raised but doctrinally weak doesn't carry the same weight as a live factual dispute. Applying that judgment here: 88–95% probability of full recovery, closer to the "essentially undisputed" band than the "genuinely disputed" one.

What actually happened

The trial court granted the landlord summary judgment for the full $520,361.29 — the Court of Appeal affirmed, holding that paying rent, not operating the gym, was the tenant's actual contractual obligation, so none of the pandemic-related defenses excused it. Read the opinion →

Reading the gap: the actual result (100% recovery) landed above even the high end of the estimated range. That's a real, instructive limit of a probability-weighted model: once a court resolves liability entirely on summary judgment, there's no partial-recovery outcome to weight against — the range approach is built for the more common case where some genuine uncertainty and settlement dynamics apply a discount, not for a clean, total win. A well-calibrated tool should show its work honestly here rather than claim a better fit than the range actually had.

The facts

A borrower/developer sued its construction lender, alleging the lender used deceptive "loan-to-own" tactics on a 25-unit condo development — failing to fund draw requests and engineering a default in order to foreclose and take the project itself. The lender countersued for $16 million.

Applying the rule

Our documented lender_liability_claim rule starts deliberately low — 15–35% probability — because lender-liability doctrine is historically borrower-unfriendly absent clear evidence of bad faith. That's not a guess about this specific case; it's a base rate reflecting how hard these claims are to win across the doctrine generally. Damages, when a claim like this does succeed, are comparable-case-informed rather than formulaic, since they depend heavily on the specific harm proven.

What actually happened

A Dallas County jury found the lender liable on every theory — fraud, breach of contract, usury, wrongful foreclosure, and promissory estoppel, with a finding of malice — awarded the developer over $6 million (growing to roughly $12.79 million at final judgment), and rejected the lender's $16 million counterclaims entirely. Read more →

Reading the gap: this is exactly the case a 15–35% probability range exists to describe — a claim type that mostly loses, but pays out enormously on the real facts that support it. A single case beating long odds doesn't mean the base rate was wrong; it means the range is honestly modeling a claim type where most matters settle for little or lose outright, and a minority with genuinely bad-faith conduct on the record can produce a verdict like this one. Judging the model by whether any one low-probability case wins or loses misses the point of a probability range — the question is whether the range is well-calibrated across many matters, which is exactly what the prediction-registry feature above is designed to eventually show.