Curve Sets (Storage Grouping)
Table of Contents
Summary
A curve set is a storage-layer grouping: several small, closely-related curve time series (one per curve in a family) persisted as one composite group rather than as separate series, because they are individually small and almost always consumed together. It is unrelated to the build-order dependency between a family's curves — grouping is a physical/storage decision, made after every constituent curve has already been built.
Detail
Why group
Persisting every curve as its own independent time series is wasteful when each series is individually small. Grouping closely-related series together reduces series count and matches how the data is actually read — callers almost always want the whole set (e.g. every curve in a currency's Funding + Projection family) rather than one curve in isolation. This generalises beyond interest rates: FX spot rates receive the same "small, related series grouped together" treatment.
Identity and keying
Every market data item, including a curve set, carries an associated reference date (evaluation/pricing date), normally chosen so the discount factor for that date is 1 (and, for other curve types, so the survival probability is 1 / integrated variance is 0). Keys describing a curve set follow the general market-data key structure: Type, Subtype, Underlyings, Fixing Date.
Relationship to the family
A Funding Curve + Projection Curves family is the natural candidate for being physically stored as a single curve set — a group of related time series: the family is the logical/financial relationship (including the build order between its curves); the curve set is the physical/storage grouping of that same set of series, decided independently of how — or in what order — those curves were constructed.
See also
- Interest Rate Curve Families — the hub.
- Funding and Projection Curves — the family typically stored as one curve set.
- Multicurve Management — how a curve set would be presented once retrieved.