Delivering Data-Driven Applications at the Speed of Business: Global Banking AML use case.
Chief Data Officers in financial services have unique challenges: they need to establish an effective data ecosystem under strict governance and regulatory requirements. They need to build the data-driven applications that enable risk and compliance initiatives to run efficiently. In this webinar, we will discuss the case of a global banking leader and the anti-money laundering solution they built on the data lake. With a single platform to aggregate structured and unstructured information essential to determine and document AML case disposition, they reduced mean time for case resolution by 75%. They have a roadmap for building over 150 data-driven applications on the same search-based data discovery platform so they can mitigate risks and seize opportunities, at the speed of business.
3. Agenda
• Introductions
• Trends in Financial Services Risk & Compliance
• Trends in the AML Space
• Why Open Enterprise Apache Hadoop for Modern Data Architectures
• Architectures & Work Streams
• An AML Case Study
• Q & A
5. Hortonworks Key Focus Areas in Financial Services
Common Focus AreasSegments of Banking
Risk Mgmt
Cyber
Security
Fraud
Detection
Predictive Analytics
Data
AML Compliance
Digital Banking
360 degree
view
Customer Service
Capital Markets
Corporate Banking and
Lending
Credit Cards &
Payment Networks
Retail Banking
Wealth & Asset
Management
Stock Exchanges &
Hedge Funds
+
6. Demand drivers for Big Data in
Retail Banking & Capital markets
Catalyst Definition Example
Larger data sets Larger data sets allow analysts to query and conduct
experiments with fewer iterations
Omnichannel data, Tickers, price, volume and
longer time horizons. Social media/ third party
data
New types of data New data types that need to be synthesized for
traditional relational databases
Business process data, Social Data, Sensor &
device data. OTC contracts and public filings.
Analytics and
visualization
More powerful analytics and visualization tools to
explain and explore patterns – Fraud, Compliance &
Segmentation
Complex Event Processing (CEP), predictive
analytics. Portfolio and risk management
dashboards
Tools and lower-cost
computing
Open source software tools. Lower server and
enterprise storage costs
Hadoop, NoSQL. Commodity hardware. Elastic
compute capacity.
7. Transformation
--- Maturity Stages à
OptimizationExplorationAwareness
---MaturityStagesà
Peer Competitive Scale
Standard among peer
group
Common among peer
group
Strategic among peer
group
New Innovations
No Use Case Name
1 Single View of Ins/tu/on
2 Predict Risk Exposures
3 Predict Counterparty Default
4
Automa/on of Client Due Diligence for
consumer onboarding
5 Enhanced Transac/on Monitoring
6 Enhance SAR Accuracy
7 Credit Risk Calcula/on
8a
Regulatory Risk Calcula/ons – Basel III &
CCAR
8b
Regulatory Risk Calcula/ons – Basel III &
CCAR
9a
Calcula/ng VaR across mul/ple trading
desks
9b
Calcula/ng VaR across mul/ple trading
desks
10
Calculate credit risks across a variety of
loan porRolios
11 Internal Surveillance of Trade Data
12
CAT (Consolidated Audit Trail)/OATS
Repor/ng
13 EDW Offload
Corporate &
IT Functions
Trading Desks
Retail Banking Use Cases are available at different levels of maturity
Surveillance
Security & Risk
2
8a
5
7
1
6
3
4
9a
10
11 12
8b
9b
13
12. General Trends in AML
Trends
• Increasing levels of criminal sophistication
• Illicit activities span geographies, products and accounts
• Expert systems and rules-engine approaches are becoming less effective
• Inefficient investigation tools and processes aren’t keeping up
Impacts for AML
• Programs must evaluate multiple, varied data sources
• Require a 360-degree view across much larger data sets
• Automated, predictive approaches must replace manual, reactive programs
13. The Current State of AML Data Analysis
• Investigators demand interactive, visually appealing user interfaces
• Data discovery and predictive analytics can show deeper customer trends
• Aging technologies and their supporting approaches should be retired
• Companies are adopting advanced risk classification approaches
• New technologies help reduce the number of “false positives”
16. What We Have Seen at Banks
Fragmented Book of Record Transaction systems
• Lending systems along geographic and business lines
• Trading systems along desk and geographic lines
Fragmented enterprise systems
• Multiple general ledgers
• Multiple Enterprise Risk Systems
• Multiple compliance systems by business line
• AML for Retail, AML for Commercial Lending, AML for Capital Markets…
• Lack of real time data processing, transaction monitoring and historical analytics
Proprietary vendor and in-house built solutions
• Acquisitions over the years have built up a significant technological debt
• Unable to keep pace with the progress of technology
• Move to combine Fraud (AML, Credit Card Fraud & InfoSec) into one platform
• Issues with flexibility, cost and scalability
18. …And the Data Complexity Continues to Grow
• Tens of point-to point feeds to
each enterprise system from each
transaction system
• Data is independently sourced,
leading to timing and data lineage
issues
• Business processes are
complicated and error-prone
• Reconciliation requires a large
effort and has significant gaps
Book of Record Transac/on Systems
Enterprise Risk, Compliance and Finance Systems
22. Leading AML Use Cases
• Large transfers across geographies
• Single view of a customer with multiple accounts
• Linked entity analysis
• Watch-list monitoring and data mining
• Credit card fraud detection
23. Major areas of activity around AML..
• Automating Due Diligence around KYC data
– Simple information collected during customer onboarding
– More complex information for certain entities
– Applying sophisticated analysis to such entities
– Automating Research across news feeds (LexisNexis, DB, TR, DJ,
Google etc)
• Efficient Case Management
• Applying Advanced Analytics (two sub Use Cases)
– Exploratory Data Science
– Advanced Transaction Intelligence
24. Stream Processing
Storm/Spark ML
Reference Architecture for Fraud/AML/Compliance
Stream
Flume
Sink to
HDFS
Transform
Dashboard
UI Framework
ELT
Hive
Storage
HDFS/Spark ML
Stream
Kafka
Stream to Kafka
Stream to
Flume
Forward to
Storm
Monitoring / KPI
NoSQL
HBase
Real-Time Index
Search
Solr
ELT
Pig
Batch Index
Alerts
Bolt to
HDFS
Dashboard
Silk
JMS
Alerts
Interactive
HiveServer
Visualization
Tableau/SAS/ETC
Reporting
BI ToolsBatch Load
High Speed Real Time
and Batch Ingest
Real-Time
Batch Interactive
Machine Learning
Improved Models
Load to
Hdfs
SOURCE DATA
Customer
Account Data/
CRM/MDM
Transaction
Data
Order
Management
Data
Click Stream
Log//Social Data
Documents
EDW
File
REST
HTTP
Streaming
RDBMS
Sqoop
JMS
33. The Advantages of Big Data AML Solutions
• Hortonworks Data Platform (HDP) is a linearly scalable platform already in
use at many of the world’s largest financial services companies
• Hortonworks takes a 100% open-source approach to Connected Data
Platforms that manage data-in-motion and data-at-rest
• Partnering with an open source vendor gives banks more options than
choosing a proprietary software platform
• Regulators are streamlining their regulatory practices by adopting a Big Data
approach
Contact Hortonworks to discuss your journey to
actionable intelligence for AML