Skip to main content
Rook compares when your selected outcomes pay. It does not infer statistical dependence from prices.

​
Analyze your selection

Send this body to POST /relationships:analyze. The result includes relationships, basket findings, per-input coverage and the scenario versions used in the comparison.

​
Pairwise and basket findings

Pairwise findings include equality, complement, implication, payout dominance, exclusivity and overlap. Each result explains which scenarios and settlement rules it covers. A basket can be classified as PARTITION, EXCLUSIVE, EXHAUSTIVE, OVERLAPPING or UNCOMPUTED. Counts distinguish no outcome paying, exactly one paying and two or more paying. Duplicate canonical outcomes are deduplicated for these set statistics. Witness references explain a gap, overlap or counterexample. witness_limit restricts returned examples, not the domain used to establish a finding.

​
Counts have a denominator

Counts depend on the published scenario version and representation. They are not probabilities. A compressed payout group must not be presented as a raw scenario count. POSITIVE_PAYOUT and FULL_PAYOUT differ when fractional or void settlements are possible. By default, incomplete coverage rejects the analysis. Explicit partial analysis describes only the supported subset; it cannot establish a conclusion for the whole selection. GET /relationships can return neighbors outside your initial market. Select those outcomes explicitly before including them in calculations. Discovery does not expand an analytical request automatically.