Why financial institutions should stop playing by the rules
Financial institutions are locked in a battle against sophisticated criminal networks. Their compliance staffs work tirelessly to do honest business, free from corruption and scandal. But they can’t keep up with malicious, global criminal networks that find new and more complex means of laundering ill-gotten gains through legitimate financial systems. Legacy, rules-only anti-money laundering (AML) systems are at their breaking point: already inefficient, they are not equipped to address these global money laundering networks’ increasing sophistication.
How a rules-only AML system works:
- Applies rules to transaction data
- Flags suspicious transactions
- Pushes these anomalies to a human investigator
Limits of rules-only AML systems:
- Don’t consider other types of data that help detect and prevent criminal activity
- Can’t understand relationships among entities
- Not able to identify larger patterns and trends
- The systems are rigid where criminals can be flexible
- Don’t take advantage of vast data stores in institutions
The best place to hide is in plain sight, and money launderers know that especially well. They deploy tactics that are difficult to detect without a holistic view of wider networks and relationships. They have broken rules-only AML systems.
Introducing graph analytics: Beat money launderers at their own game
It’s time to fight money launderers in their own territory—beyond the limits of rules-only systems. Finally, a technology has emerged which levels the playing field and exposes the complex webs of deception confounding rules-only AML platforms—that technology is graph analytics. Now is the dawn of a new era—one where financial institutions can effectively protect their organizations, reputations and customers from invasive criminal activity.
“The fight against money laundering has reached a tipping point as effective AML mitigation is becoming more challenging in an ever evolving regulatory and business ecosystem. With heavy reliance on rules-based detection and highly manual investigative processes, the financial services industry is rapidly embracing graph analytics technology. By visually connecting customers and parties, related accounts and payments, and other data, graph analytics can deliver more holistic customer profiles, uncover hidden risks, and optimize financial crime detection and investigations, while simultaneously easing the burden on staffing and elevating the customer experience.”
Graph analytics is a mathematical model that analyzes data in graph format. Data is structured as data points (or nodes) and relationships among those data points (or edges). This system allows users to connect data sets and evaluate any type of pattern or connection among entities. Because graph analytics can evaluate complex relationships and distant connections, it helps uncover and explain previously invisible patterns.
Applications of graph analytics span industries, providing insights that were unfathomable until recently. The best part for chief compliance officers? Graph analytics can arm financial institutions to defend themselves against malicious networks of money launderers.
Graph analytics from Oracle Financial Crime and Compliance Management
Oracle Financial Crime and Compliance Management (FCCM) first incorporated graph analytics in 2018, based on research from Oracle Labs. Our graph analytics capabilities are supported by our advantages in data, querying, processing, and visualizations.
Example of a graph that combines internal, external and negative news data.
1. Data at the core
- Graph analytics for FCCM is powered by our Financial Services Data Foundation (FSDF), which comprises one of the most comprehensive anti-financial crime data models in the industry. We have been refining it for over 20 years.
- Our inbuilt, proprietary Financial Crime Graph Model (FCGM) seamlessly consolidates and indexes FSDF data so that it can be analyzed and visualized using graph analytics. Flexible and easily extensible, there is no predefined schema, which is particularly useful for sparse data.
- A multimodel configuration (such as Oracle Database) provides users with the flexibility to decide how the query and manage their data.
- FCGM pulls from data lakes, relational databases, one-off datasets (in Excel, CSV, etc.) and third-party data feeds. This allows users to model connections among all their data sources, painting a bigger picture of customers and their relationships.
- Our quantified integration provides on-demand, risk-rated information based on unstructured, open source intelligence and external data sources such as trade finance documents.
- We prepackaged International Consortium of Investigative Journalists (famous for releasing the Panama Papers) data into the graph to instantly reveal and resolve potential bad actors.
2. Superior query language
- Graph query language uses logic that simplifies the expression of pattern (paths through the graph) within complex, indirect, or distant data relationships. It’s easy to use and simple to understand.
- Property Graph Query Language (PGQL) is Oracle’s proprietary, SQL-like language. It expresses queries in a compact way, which makes coding and processing easier and faster. PGQL is an open-sourced project led by Oracle.
- PGQL queries can be processed one to two orders of magnitude faster than similar queries run in SQL on tabular data.
“With graphs, data can be managed in more intuitive ways, closer to how people organize their thoughts on a white board. Our system takes advantage of parallel processing and the huge amounts of memory available in modern servers. This allows us to directly model the relationships among all of our data.”
—Hassan Chafi, Senior Director, Research and Advanced Development at Oracle Labs
3. Faster processing
- Oracle uses a specialized, highly scalable, in-memory graph analytics engine called Oracle Parallel Graph Analytics, or Oracle PGX.
- Parallel processing and huge amounts of available memory allow for lightning-fast response.
- Built-in, PhD-level algorithms let users quickly perform some of the most common queries. Users can customize these algorithms and modify constraints to suit their needs.
- An API allows users to build their own algorithms, improving accuracy for those who want complete customization.
4. Powerful visualizations
- Allow users to intuitively explore relationships among all sorts of data points by simply clicking from one node to the next.
- Graph analytics visualizations, powered by open source data science notebooks, are embedded in our Enterprise Case Management (ECM) application via the Investigation Hub module.
- Each visualization is instantly updated in real time. The visualizations are intuitive and help provide context for risk scores.
- Once a data scientist can visualize connections, it’s easier to create an algorithm. Having graph analytics as part of a machine learning ecosystem, therefore, improves accuracy.