Struct RiskFrame 

pub struct RiskFrame<'a> {
    risk: &'a RiskVector,
    npv: f64,
    market: &'a MarketContext,
}

A risk vector, the value it came from, and the market that explains both.

Holds the market by reference, so a frame built per lens-filtered sub-book borrows one set of ladders rather than copying a matrix per book.

Fields

risk: &'a RiskVector

npv: f64

market: &'a MarketContext

Implementations

impl<'a> RiskFrame<'a>

pub fn new(risk: &'a RiskVector, npv: f64, market: &'a MarketContext) -> Self

A frame from its parts, so a risk vector collected outside a book valuation is measured through the same path as one collected inside.

pub fn npv(&self) -> f64

The value the pass produced.

pub fn risk(&self) -> &'a RiskVector

The vector as collected — per parameter, unweighted and unflipped.

pub fn valuation_date(&self) -> Option<NaiveDate>

The date tenor buckets are measured from: the reference date the market’s curves share.

pub fn read( &self, axis: RiskAxis, basis: Basis, aggregation: Aggregation, scale: Scale, ) -> Result<Vec<(Label, f64)>, RiskError>

One reading of the frame: an axis, a basis, an aggregation, and the scale the rows are reported in.

The scale is applied here and nowhere else, so a report and a frame read at the same scale are the same numbers.

Errors

RiskError when a Basis::Par reading meets a curve pinned without its ladder, or when a Aggregation::Bucketed reading has no valuation date to measure tenors from.

fn read_unit( &self, axis: RiskAxis, basis: Basis, aggregation: Aggregation, ) -> Result<Vec<(Label, f64)>, RiskError>

read per unit quote.

pub fn read_buckets( &self, axis: RiskAxis, basis: Basis, edges: &[Period], scale: Scale, ) -> Result<Vec<(Label, f64)>, RiskError>

A zero or par reading on axis bucketed on edges rather than the fixed ladder: a bucket up to each edge, measured in years from the valuation date as the fixed ladder is, then one past the last.

Errors

RiskError::BucketEdges for no edges, or edges that are not day, week, month or year tenors, positive and rising; otherwise as read.

fn bucket( &self, contributions: Vec<Contribution>, edges: &[f64], ) -> Result<BTreeMap<usize, f64>, RiskError>

contributions summed by the first of edges, in years from the valuation date, each one’s date falls on or before; the last edge is open-ended.

fn zero_contributions( &self, axis: RiskAxis, parallel: bool, ) -> Vec<Contribution>

The axis’s entries, each weighted into its own curve node’s quote.

parallel reads each entry as its share of the axis moving as a whole, for a sum: a fitted curve’s shape parameters and a SABR node’s ρ and ν then weigh nothing, where a ladder shows each as its own row.

fn par_contributions( &self, axis: RiskAxis, ) -> Result<Vec<Contribution>, RiskError>

The axis’s entries flipped onto the instruments that calibrated them.

∂V/∂q = Σ_p (∂V/∂p)·(∂p/∂q) where p runs over the pillars of the ladder a rung belongs to — which is the Jacobian’s column, read down. The contraction takes the raw parameter sensitivity, not the zero-basis one: ∂z/∂q is already expressed against the parameters the curve stores, so weighting first would apply ∂p/∂q twice.

Columns are keyed by the ladder’s identity and position, not by the role the curve was pinned under. A curve serving two roles shares one ladder, so both roles’ pillars land on the same column and sum there — which is the right answer for an instrument that calibrated a curve doing two jobs, and is not reachable if columns are keyed by role.

On the vol axis the same flip runs through a SABR fit: a node’s sensitivity lands on the grid quotes its slice was fitted to, through the fit’s quote Jacobian. A grid node already is a quote and stands as itself.

Trait Implementations

impl<'a> Clone for RiskFrame<'a>

fn clone(&self) -> RiskFrame<'a>

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl<'a> Copy for RiskFrame<'a>

impl<'a> Debug for RiskFrame<'a>

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.