Find the scenario your portfolio is not ready for.

The proposed workflow is direct: describe the portfolio and a candidate action — hold, rebalance, move to cash, or a constrained hedge — then branch the scenarios that would hurt, shear the assumptions, and compare actions across branches. Synthetic Scenario Branches are disclosed and labelled. Any future use of permissioned or confidential data begins only after applicable data-rights, privacy, security, and legal terms are established. This is a research prototype inside a design-partner program, not a validated trading product.

Each case reports the measures it declares: expected modeled loss, CVaR tail loss, concentration, turnover and implementation cost, constraint failures, rank stability, and regret where a valid evaluation exists. A case may state its objective in prose or as expected loss + λ × CVaR + μ × implementation cost; the weights are declared per case and have not been calibrated. The output is a Shear Profile — how each action degrades as assumptions shear — because a single universal number would need a normalization and calibration protocol that does not exist yet.

The intended output identifies the modeled failure mode, the action that holds up best within the declared branch family, its trigger, principal downside, reversal condition, evidence tier, and provenance. A human with the mandate decides what to do; the prototype does not execute.

Where we publish an example, it may compare hold, rebalance, cash, and constrained-hedge actions across disclosed synthetic branches, or future permissioned branches after the required terms exist. Every example carries its provenance label wherever it appears. Examples show the shape of the analysis, not a track record.

The prototype analyses cases and reports where they break. It does not act on them.