In July, the FinTech Open Source Foundation (FINOS) released CDM 7 on GitHub consolidating more than 130 individual upgrades and development releases and followed up with a detailed analysis on August 14. The analysis was authored by Lionel Smith-Gordon and Ben Page of REGnosys, an active corporate member and key technology contributor to FINOS.
The CDM is a machine-readable and machine-executable representation of financial products, trades, lifecycle events and business processes. Over the past eight years, it has developed from a collaboration between the International Swaps and Derivatives Association (ISDA) and REGnosys.
The first machine-readable version covered simple interest-rate and credit derivatives and a defined set of business events. ISDA continued developing derivatives coverage while the International Capital Market Association (ICMA) and International Securities Lending Association (ISLA) extended the model into bonds, repo and securities lending. In August 2021, the three associations agreed to coordinate its future development. They selected FINOS in September 2022 to host the repository, expand the contributor community and support wider adoption while the associations retained oversight of contributions and the relevant market-specific standards.Functional Ingestion Mappings
The most significant change is functional ingestion. Earlier implementations used synonym mappings and specialist Java code to convert Financial products Markup Language (FpML) confirmations into CDM objects. CDM 7 expresses that logic as reviewable Rune functions and uses a flatter FpML representation closer to the source documents. The analysis authors call these “functional ingestion mappings”. Coverage extends across rates, foreign exchange, credit, equity and commodity products.
Ingestion is where a common model meets a firm’s internal data. Clearer logic may help business analysts and data specialists follow how a source field becomes a CDM object. It also makes mappings easier to review and change across reporting and post-trade processes. Firms using proprietary booking models must still map their internal trade and lifecycle data into CDM, but CDM 7 makes those transformation rules easier to inspect, test and maintain
CDM 7 connects legal agreements more closely to the transactions and events they govern. The release migrates material from ISDA Foundations and extends support for Credit Support Annexes, Credit Terms Annexes, Global Master Securities Lending Agreements and umbrella agreements. The model also captures elections, calculation-agent provisions and valuation responsibilities. This provides common components for derivatives, collateral and securities-finance processing.Executable workflows on Shared Structures
CDM 7 introduces the first major components of an Instruction Composition Framework, beginning with interest-rate reset processing. The authors say this creates a foundation for “more deterministic, auditable and reusable lifecycle workflows”
Changes to the Asset, Product and Observable models reduce duplication and modelling complexity. CDM 7 aligns asset classification, observable modelling, scheduled and contingent transfers, qualification functions and temporal concepts. Commodity and equity taxonomies receive further work. These shared structures allow a wider range of products and processes to be represented without creating separate models for each market.
The Event Model receives a similar clean-up. Functions for transfers, resets, quantity changes, terminations and securities-finance events have been refactored. Calculations move closer to instruction generation, making processing decisions more explicit. Operations, reporting and risk systems could therefore work from a more consistent account of the same lifecycle event.
Collateral, margin and securities finance gain wider coverage. The release develops collateral eligibility, inventory and locate workflows, debt and equity classifications, guarantor details and securities-lending events. It also improves support for standardised initial-margin calculations.
Shared Definitions Strengthen Controls
Collateral and securities-lending processes can share more product, agreement and event components with derivatives, instead of remaining in separate operational models.
Reference-data controls have been tightened. CDM 7 updates FpML coding schemes, International Organization for Standardization currency data, Legal Entity Identifier support and taxonomy classifications. Validation tests compare CDM enumerations with the external standards they represent. This is fairly unglamorous work, but a common process cannot produce consistent results when connected systems define currencies, entities or products differently.
Rune DSL 10 strengthens the language used to define the model and its executable functions. It applies stricter validation and modelling conventions, resolves cardinality and metadata problems, and removes legacy syntax and duplicated structures. Errors should surface earlier, while contributors have clearer rules when extending CDM to another product or process.
The supporting infrastructure has also been updated. FINOS reports faster builds, revised dependencies, consistent Java namespaces, schema-publication support and the removal of unsupported generators. None of these changes adds market coverage directly. They reduce technical debt and may make later releases easier for firms and vendors to absorb.
Wider Developer Access & Testing
Sample data, testing and visualisation have expanded. CDM 7 adds ingestion tests, legal-agreement samples, securities-lending visualisations, improved validation logs and regression-maintenance tools. Firms can examine how the model behaves before applying it to production data. Contributors also gain a firmer baseline for testing changes across asset classes and deployment environments.
A new JSON serialisation format standardises how CDM objects are written and read. This should improve exchange between implementations, although existing CDM 6 data must be converted. Persisted objects, interfaces and test packs may all be affected, so firms cannot treat the upgrade as a simple library replacement.
Python distribution through the Python Package Index (PyPI) opens the model up to the wider developer community. Data engineers, quantitative teams, analysts and machine-learning developers can use generated CDM classes and supported Rune-defined functions in familiar Python environments.
CDM 7 increases the coverage of rates, foreign exchange, credit, equities, commodities, collateral and securities finance, promising to reduce the effort needed to expand model adoption for asset classes and business processes. The wider gains are more transparent mappings, stronger interoperability, clearer audit trails and less duplicated processing logic. However, no model is a silver bullet. Firms must first convert data, update integrations and retest implementation workflows before those gains become available.
Subscribe to our newsletter
