Risk, portfolios and historical snapshots

Evaluation requests follow the Rust library’s Request and EvaluationResults. Direct evaluation returns one number or table. Use a stored result when several requested measures share one evaluation:

H2 =QS.EVALUATE.RESULTS(context,trade,{"Value","Cashflows","Sensitivities"},pricer)
H3 =QS.RESULT.VALUE(H2)
H4 =QS.RESULT.SENSITIVITIES(H2)
H5 =QS.RESULT.COUPONS(H2)
H6 =QS.RESULT.CASHFLOW.SUMMARY(H2)

Sensitivities and cashflows

Sensitivities are raw derivatives in the pricer’s value units per unit factor. For a decimal-rate factor, multiply by 0.0001 for an approximate +1 bp value change. Retain each factor’s actual units; futures price points, currency spots, volatility and rates are not interchangeable. Risk is calculated from the original trade and model, not from a compressed cashflow report.

Cashflows contain signed undiscounted amounts in each payment currency, with positive receipts and negative payments. The detailed coupon report retains accrual dates, fixing and contractual indices, notional and rate conventions. The summary groups Interest, Capital, Other and Total by payment date/currency. It does not convert currencies or allocate accrual-period net interest income.

FairRate(index) requests one zero-based leg or loan coupon segment. Swap leg 0 solves the fixed coupon and leg 1 solves the floating spread, including all legs and principal in the zero-PV condition. Loan segments are solved successively, locking earlier rates. Bounded loan coupons remain unsupported for fair-rate solving. Separate targets require separate evaluations.

=QS.EVALUATE(context,trade,"FairRate(0)",pricer)
=QS.EVALUATE(context,trade,"FairRate(1)",pricer)
=QS.REQUEST.FAIR.RATE(0)

Portfolio aggregation

B2 =QS.PORTFOLIO.CREATE("book",A10:A100)
B3 =QS.EVALUATE.PORTFOLIO(context,B2,pricer,requests,"USD")
B4 =QS.RESULT.VALUE(B3)
B5 =QS.RESULT.COUPONS(B3)
B6 =QS.RESULT.CASHFLOW.SUMMARY(B3)

The portfolio uses the caller-supplied name as its identifier and holds an ordered trade-reference array. Repeated handles count as repeated positions. Pass one pricer for all members or one per member. Value and Sensitivities are computed natively in the context base currency, including FX risk. The optional currency argument must match that base currency. Unsupported or incomplete measures remain absent.

Evaluation streams one trade at a time and releases its intermediate result. Compatible coupons are grouped using notional-weighted rates or spreads, while pay/receive directions, nonlinear payoffs and distinct conventions remain separate. The portfolio does not retain a combined AD graph. For individual scalar values, use EVALUATE.RANGE with one shared committed context instead.

Archive complete dated markets

Prepare every root with its full contribution array. Put prepared-context references in A2:A10 and corresponding observation timestamps in F2:F10:

C2 =first_root
C3 =next_root
H2 =QS.SNAPSHOT.FROM.RANGE("history",A2:A10,TRANSPOSE(F2:F10))
H3 =QS.EVALUATE.FROM.SNAPSHOT(H2,trades,dates,pricer,"Value")

Observation timestamps require an explicit RFC 3339 offset, for example 2026-01-15T16:00:00Z. The selected quote-store reference date remains the actual valuation date, even if an observation crosses a UTC day boundary. Dates and trade references align one-for-one; blanks are rejected instead of shifting rows. The output follows the original input order.

Defaults are Previous/Mid. Previous selects the most recent complete observation at or before a request; Exact requires an exact match; Closest may select a future observation and chooses the earlier one on a tie. Missing Bid/Ask never falls back to Mid. Duplicate UTC observation timestamps fail explicitly.

For individual selection and inspection:

J2 =QS.SNAPSHOT.SELECTOR("selection","2026-01-16T17:00:00Z","Previous","Mid")
J3 =QS.SNAPSHOT.SELECT("matched",H2,J2)
J4 =QS.SNAPSHOT.TIMESTAMP(J3)
J5 =QS.SNAPSHOT.REFERENCE.DATE(J3)
J6 =QS.CONTEXT.FROM.SNAPSHOT("historical",H2,J2,fixings)

Snapshots store quote observations, FX and construction configurations, including direct zero-rate quotes, credit, volatility and simulation settings. Supplied runtime curves cannot be captured. Fixings and their publication vintages are separate; supply a fixing contribution or committed context to historical pricing. No fixing-vintage filter is inferred. Native JSON import/export is available under SNAPSHOT.FROM.JSON and SNAPSHOT.TO.JSON; Excel limits exported cell text to 32767 UTF-16 units.

Historical batch performance

The native batch groups rows by matched snapshot and constructs each market once. It returns scalar values in the original input order without creating per-row context or evaluation-result objects. Evaluation completes synchronously during Excel calculation.

The historical and context example workbooks show this batch API. Portable tests verify snapshot selection, calibration reuse, input ordering and stale-handle validation. Worksheet execution requires Windows Excel.