8 PRODUCT ARCHITECTURE OVERVIEW
8.1 Introduction
FpML incorporates a significant level of structure, rather than being a 'flat' representation of data. This structuring is achieved through the grouping of related elements describing particular features of a trade into components. Components can both contain, and be contained by, other components.
An alternative approach would have been to collect all the required elements in a single large component representing a product or trade. A flat structure of this kind would capture all the relevant information concisely but would also constrain the model in two important respects, namely, ease of implementation and extensibility.
Grouping related elements into components makes it easier to validate that the model is correct, that it is complete and that it doesn't contain redundancy. This is true, both from the perspective of readability to the human eye, and also from the perspective of processing services. Processing services that do not need all the information in a trade definition can isolate components and be sure that the complete set of elements required, and only the elements required, is available for the particular process in hand.
Components additionally serve as the building blocks for a flexible and extensible model. Generally speaking, the complexity of financial products is a result of combining a few simple ideas in a variety of different ways. The component structure supports a trade content definition that is flexible enough to represent the wide variation of features found in traded financial instruments.
It should be noted that the application of the guiding principles of extensibility and ease of use has resulted in a different approach with regard to the forward rate agreement. Because this product is straightforward, commoditized and unlikely to develop further, the advantage to be gained from the extensive use of components is outweighed by the concision of a single component.
8.2 Component Framework
The optimum level of granularity is important to FpML. FpML separates the elements which collectively describe a feature of a product or trade into a separate component with each component serving a particular semantic purpose. Every grouping of elements in FpML is regarded as a component and each component is regarded as a container for the elements that describe that component. In the majority of cases each component will contain a mixture of other components and primitive elements, e.g. a date or string, that collectively describe the features of the component. Components are typically represented in the FpML Document Type Definition (DTD) as XML entities.
Generally speaking, the lower level a component is, the more re-usable it will be. FpML makes use of a number of primitive entity components that describe the basic building blocks of financial products, for example, FpML_Money, FpML_AdjustableDate, FpML_BusinessCenters, FpML_Interval, FpML_BusinessDayAdjustments etc. These primitive components are re-used in different business contexts.
Primitive components are contained in higher level components that describe the features of particular products. For this reason these higher level components will tend not to be re-usable to the same extent. Examples within the definition of swapStream are the components required to construct schedules of dates such as calculationPeriodDates, resetDates and paymentDates. However, it should not be inferred from this that any fundamental distinction is drawn between components in usage or structure.
8.3 Overview of Core Components
The root element contains three elements, trades, portfolios and parties. Portfolios contain only trade references, if the trade itself needs to be included in the document then the trades can be included within the root element.
8.3.1 The Trade Component
The trade is the top-level component within the root element FpML. A trade is an agreement between two parties to enter into a financial contract and the trade component in FpML contains the economic information necessary to execute and confirm that trade. A trade contains four components: tradeHeader, product (an abstract concept), party (two or more instances) and otherPartyPayment (zero or more instances).
8.3.1.1 tradeHeader
The information within tradeHeader will be common across all types of trade regardless of product. In FpML 2.0 this contains the trade date, party trade identifiers and any calculation agent references.
8.3.1.2 product
Product is an abstract concept in FpML and an actual product element is not used. Instead, one of the FpML products will appear directly under trade.
8.3.1.3 otherPartyPayment
This component contains additional payments such as brokerage paid to third parties which are not part of the economics of a trade itself.
8.3.2 The Portfolio Component
The portfolio component specifies a set of trades as a list of tradeIds and a list of sub portfolios. Portfolios can be composed of other portfolios using a composition pattern. By using the tradeId to identify the trade the standard allows for portfolios to be sent around without the full trade record.
8.3.3 The Party Component
The party component holds information about a party involved any of the trades or portfolios included in the document. The parties involved will be the principals to a trade and potentially additional third parties such as a broker. For this release, this component is restricted to party identification.
It should be noted that an FpML document is not 'written' from the perspective of one particular party, i.e. it is symmetrical with respect to the principal parties. The particular role that a party plays in the trade, e.g. buyer, seller, stream payer/receiver, fee payer/receiver, is modeled via the use of references from the component where the role is identified to the party component.
8.3.4 The Product Component
The product component specifies the financial instrument being traded. This component captures the economic details of the trade. Because of the complexity of the OTC Interest Rate Derivatives domain that FpML 2.0 covers, composing these products from various building blocks is a key aspect of the design approach.
FpML 1.0 focused on the instrument definitions for interest rate swaps (including cross currency swaps) and forward rate agreements. For that initial release, a trade was restricted to containing only a single product definition. In FpML 2.0 product strategy has been introduced which allows more than one product within a trade. In FpML 2.0 the instrument definition has been extended to include options.
8.3.5 The Strategy Component
This component allows the structuring of trade by combining any number of products within a strategy. This component makes use of a composition pattern since strategy itself is a product. This means that strategies can themselves contain strategies.
8.4 Coding Schemes
A necessary feature of a portable data standard is both an agreed set of elements and an agreed set of permissible values (the value domain) for those elements. An FpML document exchanged between two parties would not be mutually understandable if either or both of the parties used internal or proprietary coding schemes to populate elements.
Reference data can originate from various sources and the range of permitted values may be more or less extensive. The dayCountFraction is an example of an element with a limited set of permissible values with well-defined meanings. The range of permitted values comes from several sources including ISDA and AFB definitions. However, the currency element is an example of where the list of permitted values is more extensive and the coding scheme reference is to a well-known standard, in this case ISO 4217.
In FpML 1.0 the recommended domain for party identification is a valid bank identifier code (BIC). The BIC is an ISO standard, ISO 9362. S.W.I.F.T. is the designated registration authority for the assignment of BIC codes.
One possible means of identifying value domains would have been to include the domain of permitted values within the DTD. This solution has been rejected for two reasons. Firstly, in many cases the scope of permitted values is extensive, most obviously with party identifiers, and this would make the standard unnecessarily bulky. Secondly, although there are varying degrees of stability, all value domains are subject to change and including them in the DTD would have necessitated a new version of FpML each time a value domain changed.
For these reasons, FpML uses Schemes to identify the permitted values for an element. In each case, the reference Scheme will be identified by a URI. The URI will either identify a well-known external standard such as ISO 4217, or where no well-established standard exists, an FpML standard. FpML 1.0 includes provision for a default Scheme and the facility to override the default Scheme at an element level. In both cases, no values are included for the URI in the DTD in order to avoid coding either particular Schemes, or particular versions of Schemes, into FpML. For the same reason, the URI quoted in an FpML document for a Scheme that is FpML controlled will include a date and version in order to identify the particular version referenced.
It should be noted that the Scheme approach adopted by FpML does not allow validation of the values within the DTD. It will be the responsibility of the applications that implement FpML to validate that the contents of an element conform to the specified Scheme.
For further details on the architectural framework behind Schemes, refer to the FpML Architecture Version 1.0 document.
