Realized Outcome

A realized outcome is what a matter actually resolved to, captured as a structured label rather than a narrative summary, and matched back to the prediction made about that matter before resolution. Capturing realized outcomes is what closes the loop between prediction and evidence: without it, a model's calibration claim cannot be checked against reality, and a claim of accuracy is unverifiable rather than merely unverified. A distinct and stricter meaning of the term applies to claims about realized-dollar returns specifically, the actual capital-side proceeds a funded position produced, as opposed to a labeled settlement or award distribution built from court and regulatory records. This platform does not claim that a current model predicts a realized-dollar return for any given asset class until a model trained on counterparty tapes, actual realized-dollar records from capital partners in that specific asset class, has passed a preregistered backtest designed and locked before the results were seen. Activation of this claim happens one asset class at a time, because the mechanism converting a labeled distribution into realized capital-side dollars differs meaningfully across asset classes, a mass tort settlement, a single-plaintiff advance, an insurance subrogation recovery, and evidence in one class does not transfer to another. Absent that specific, class-by-class evidence, the claim remains inactive regardless of how well the platform's outcome-probability models otherwise perform.