Application Development
Cyber Security
Featured
Management Consultancy
RA
Author
Roua Achi
Read time
6 minute read
Published
23 Jun 2020
Topics
AI and Data Science
What is Master Data Management?
One golden source — across every system, every process.
Master data is the core data needed to uniquely define objects like parties (customers, vendors, suppliers, trading partners, or employees), places (locations or geographies), and things (products, services, or accounts). It does not change as frequently as transactional data and is referenced by business processes and other applications. This data is usually located in multiple applications and is often out of synchronisation, without a true “golden” source.
MDM is the tool to ensure that an organisation has only one version of the truth — providing the 360-degree view of data that is of interest to the business.
MDM vs Reference Data
Are they the same? Not quite.
Master Data and Reference Data are related but distinct. Understanding the difference is foundational to building the right data architecture for a financial services organisation.
Why do I need MDM?
I already have a data warehouse — why add MDM?
If you already have a data warehouse, you are in a very good place to start. But most data warehouse ecosystems have attempted to manage master data within the warehouse architecture — typically focusing on mastering data after transactions occur. This approach does little to improve data quality because data is corrected after the fact.
The best way to improve data quality is to move the process upstream of the data warehouse — before transactions are executed. MDM can make your organisation smarter and more flexible: by having accurate data for your most important information, you can be sure that your models, projections, and predictions are as accurate as they could be.
Automating trade executions
Fast trade execution is important to brokers because it allows them to capitalise on rapidly changing market opportunities. Accurate master data is the foundation of every automated execution workflow.
Supporting compliance
Compliance officers need to report on existing regulations and dynamically respond to new mandates at the national and state level (such as Basel II). They require desktop access to reconciled data within and across customer, counterparty, and financial instrument domains.
Managing risk
Having a clear picture of an institution’s holdings is essential for accurately assessing and adjusting risk levels. Bankers require desktop access to reconciled and related data within and across customer, counterparty, and financial instrument data domains.
Reference data lifecycle
Four stages — from source to distribution
The goal of a data management strategy is to develop the operational platform to support a world-class hedge fund, minimise operational inefficiencies, errors and costs, and reduce legal, regulatory, and operational risk. The reference data lifecycle defines how data flows from raw acquisition through to real-time distribution across consuming systems.
Reference data lifecycle — Acquire → Cleanse → Maintain → Distribute
1
Acquire
→Internal & external sources
→Extract, Transform, Load
→
2
Cleanse
→Business rules
→Golden copy validation
→Consolidations
→
3
Maintain
→Security & issuer setups
→Data quality
→Governance review
→
4
Distribute
→SLA-governed publish
→Workflow & rules
→Real-time to key systems
Goal: to develop the operational platform that supports a world-class hedge fund — minimising operational inefficiencies, errors, and costs while reducing legal, regulatory, and operational risk.
Implementation
MDM quick implementation checklist
Each step requires careful consideration — but starting with a clear sequence ensures the initiative builds momentum without overreaching in scope.
Challenges
Two common challenges in implementing MDM
Getting business involvement
MDM must be driven by business needs — not IT
MDM has to be driven by business needs, otherwise it could turn out to be just another database that needs to be synchronised with all others — making it more of a liability than an asset. The difference between MDM success and failure depends greatly on an organisation’s ability to determine its own definition of what constitutes quality, trustworthy data. Most hedge funds already have concepts such as a Pricing Community which can act as Data Stewards — liaisons between business and IT who facilitate discussions about data and determine MDM requirements.
Big vision, baby steps
Limit initial scope — then expand
Consider the ultimate goal, but limit the scope of the initial deployment. Once master data management is working in one domain, it can be extended to others. An important differentiator: each MDM implementation should include domains that work together — such as Customers and Products — rather than attempting to master all data at once.
Implementation options
Trends in reference data implementation — four paths
Common architectural trends include a drive towards multi-asset trading platforms with consolidated infrastructure — requiring real-time data distribution for trading, risk management, and compliance. These four implementation options represent the spectrum from full custom build to full outsourcing.
Option 1 — Buy/build custom
3rd party feed handling and transformation, with custom consolidation, maintenance, and distribution using a common API. Maximum control, highest development effort.
Option 2 — 3rd party + custom distribution
Adopt a 3rd party product to handle all functions up to distribution, with a custom translation layer to a common API. Balances vendor capability with flexibility.
Option 3 — Full 3rd party product
Adopt a 3rd party product that includes all functions including distribution. Fastest time to deployment; lowest internal resource requirement.
Option 4 — Outsource
Outsource some or all functions — for example, allowing a 3rd party to conduct multiple-source consolidation. Reduces internal operational burden while leveraging specialist data providers.
Architecture design
Centralized vs. distributed — and the value of centralizing
The key to success when implementing an enterprise reference data solution is to balance business needs with operational efficiency. The choice between centralized and distributed architecture shapes every downstream data quality, governance, and cost outcome.
Centralized architecture
→Business rules
→Identifier assignment
→Common data standards
→Service level agreements
→Data residence
Distributed architecture
→Data capture and maintenance
→Ownership of data
→Enrichment (prioritisation and consolidation)
→Quality assurance and acceptance
Centralized reference data value proposition
Better data quality
Enterprise-wide SMF definitions improve interoperability across the portfolio management cycle. A golden copy with a centralized validation engine and best-of-breed composite record.
Lower cost of doing business
Fewer touch-points, more efficient maintenance processes, reduction in Bloomberg terminals, and a more efficient data acquisition process.
More efficient trade processing
Less manual review of new security set-ups, improving accuracy and reducing failed trade rates across all asset classes.
Common organizational trends
Centralization of data service provisioning, single point of compliance and audit for enterprise data, data stewardship roles across business and IT, and reference data outsourcing.
One golden source. Complete data confidence.
OmniVista · AI & Data Science · Management consultancy
Building a master or reference data management strategy for your fund?
OmniVista provides data strategy, architecture design, and implementation support for investment managers, hedge funds, and financial services organisations.
