portfolio

SuperLSS.portfolio(

X: object | EagerFrame,

*,

sample_weight: NDArray | None = None,

offsets: Mapping[str, NDArray] | None = None,

**kwargs: Any,

) → Portfolio

Return the simulated total over a book of rows, optionally by segment.

Each row is simulated on its own predictive law and the draws are summed across rows, so the reported quantiles are of the book total and carry the dependence the shared coefficient draws induce. sample_weight enters twice under prior semantics – every row is simulated on its own prior-weighted law and what the book pays is sum w y – and is refused under frequency semantics, which has no row-law slot.