Domino and AWS: collaborative analytics and model governance at financial services firms
1. Domino and AWS: Collaborative Analytics and
Model Governance at Financial Services Firms
2. Introduction
Financial Services Industry
• Highly regulated
• Broad range of firm sizes, from small hedge funds to global
money center banks
• Large amount of analytical work related to transactions (e.g.
mergers and acquisitions, securitizations, equity IPOs)
• Subscribers to multiple data vendors
• Historically have been massive consumers of technology
Technical Challenges
• Large amounts of legacy analytics
• Subject matter experts technically proficient, but not software
engineers
• Collaborative analytics between front office desks, risk
departments and back office can get complex
• Desktop environment and version management creates
challenges for reproducibility and collaboration
• Multiple version control file storage and database systems
• Spreadsheets and shared drives that never stop growing
3. Financial Service Industry Analytics
Common Tasks
• Building predictive risk models
• Validating the predictive models (often by a different group)
• Derivative pricing
• Constructing and back testing algorithmic trading strategies
• Credit risk stress testing
• Transaction analytics
• Reconciliation / tie-out
• Data driven research
4. Financial Service Industry Analytics
Typical quantitative developer profile
• Masters/pHd in math, physics or statistics
• Deep subject matter expertise in a rather narrow specialty
• Often build and support their own analytical stack. An
example:
• High powered local machine
• Sas, R, Matlab, SPSS
• C++
• SQL
• Excel, csv
• Maintain large libraries themselves
• Some (not all) can be a bit reluctant to adopt new technologies
into their stack
5. DBRS Overview
DBRS is an internationally recognized credit rating agency that has been providing issuers,
regulators, investors and intermediaries with objective, transparent, insightful risk analysis and
opinion since 1976.
DBRS rates entities in:
Argentina
Australia
Austria
Barbados
Belgium
Brazil
Canada
Cayman Islands
Chile
China
Colombia
Cyprus
Denmark
Finland
France
Germany
Greece
India
Ireland
Italy
Japan
Luxembourg
Malta
Mexico
Netherlands
New Zealand
Norway
Peru
Portugal
Spain
Sweden
Switzerland
Turkey
UK
USA
Uruguay
DBRS employs approximately 440 professionals around the world, with offices in New
York, Chicago, Toronto, London and Mexico City.
Public Finance
(includes Sovereigns) – 39%
Corporates – 6%
Financial Institutions &
Insurance – 26%
Structured Finance – 29%
Coverage by
Industry
DBRS Overview
6. DBRS & Domino
Securitization Analysis Example – Residential Mortgage Backed Security
Investment
Decision
Predictive
Analytics
Train/test regression
model on large historical
mortgage data set
Transaction
Analytics
For a given default &
interest rate scenario,
forecast the interest and
principal paid to the bonds Qualitative
Analysis
Legal documentation
review (true sale, events
of default)
Traditional data
science (R, Python)
Fixed income
analysis (Excel,
VBA, Python)
Human group decision
making (Word,
PowerPoint, Email)
7. DBRS & Domino
Traditional Approach Modern Approach
Data exploration / wrangling
• Excel
Scripting
• VBA
Technical computing
• Stata, Matlab, R
Storage
• Shared drive, SQL Server
Model engines
• C#, VBA
Front end
• Excel add-ins
Presentation
• Powerpoint
Data exploration / wrangling
• Jupyter notebooks
Scripting
• Python
Technical computing
• Python (numpy, pandas, sci-kit), R
Storage
• S3, Athena, Domino projects
Model engines
• Python
Front end
• Excel add-ins, R Shiny
Presentation
• Plotly, Jupyter notebooks, Powerpoint*
*Still can’t escape it
8. DBRS & Domino
Model Governance Example
Development
• Parameter estimation performed in Jupyter notebooks, which
are easily shared with the model review team
• Excel add-in which allows developers dump transaction data to
json files, Domino via REST API, or S3 bucket
• Developer codes locally with their IDE, using json files to debug
• Code changes reviewed and approved via GitHub
• Domino project automatically synced to the repo
• Developer can easily setup the correct environment for their
project inside of Domino
Advantages
• Analysts interact with Excel/R Shiny front end as they normally
would, the code is all executed server side on Domino
• Analyst workflow is auditable in the form of a Domino run,
which is a record of the code version, inputs, outputs and
environment
• Scalable via AWS, large batch jobs are easily scheduled
• Developers can control their code base by hosting it server side
without any knowledge of web technology, Docker containers,
etc.
• Development can still be very rapid & flexible
15. Conclusions
Domino & AWS
• There is a large population of analysts who are “part time” data scientists, and are not using the optimal technology
stack. Collaborative analytical solutions enable them to get up to speed rapidly and help reduce silos
• Domino provides easy access to AWS, git, docker containers, and REST API’s for non-software engineers
• There is strong momentum for AWS, not only for large firms but small ones as well
• While the modular nature of the Python/R ecosystems provide many benefits, one must be cautious of environment,
especially package management
• Subject matter experts should focus on business logic & model design, let somebody else do the plumbing!
Editor's Notes
AWS serves hundreds of thousands of customers in more than 190 countries.
Amazon CloudFront and Amazon Route 53 services are offered at AWS Edge Locations
AWS serves hundreds of thousands of customers in more than 190 countries.
Amazon CloudFront and Amazon Route 53 services are offered at AWS Edge Locations
AWS serves hundreds of thousands of customers in more than 190 countries.
Amazon CloudFront and Amazon Route 53 services are offered at AWS Edge Locations
AWS serves hundreds of thousands of customers in more than 190 countries.
Amazon CloudFront and Amazon Route 53 services are offered at AWS Edge Locations
AWS serves hundreds of thousands of customers in more than 190 countries.
Amazon CloudFront and Amazon Route 53 services are offered at AWS Edge Locations
AWS serves hundreds of thousands of customers in more than 190 countries.
Amazon CloudFront and Amazon Route 53 services are offered at AWS Edge Locations
AWS serves hundreds of thousands of customers in more than 190 countries.
Amazon CloudFront and Amazon Route 53 services are offered at AWS Edge Locations
TALKING POINTS:
WE SPEND A LOT OF TIME TALKING TO OUR FSI CUSTOMERS AROUND THE WORLD. OUR CUSTOMERS TELL US:
REGULATORY COMPLIANCE CONTINUES TO BE A SIGNIFICANT EXPENSE DRIVER, AS REGULATORY OVERSIGHT CONTINUES TO EXPAND
COMPETITION – FROM NEW FINTECH ENTRANTS AND TRADITIONAL PLAYERS – IS AT AN ALL TIME HIGH; FINANCIAL CONSUMERS HAVE MORE CHOICE NOW THAN AT ANY OTHER TIME IN HISTORY
SECURITY IS JOB ZERO FOR FSI FIRMS
FIRMS ARE SITTING ON PETABYTES OF DATA – EITHER FOR COMPLIANCE PURPOSES OR BECAUSE THEY THINK THAT IF THEY CAN MINE THAT DATA, THEY’LL BE ABLE TO IDENTIFY ACTIONABLE INFORMATION
CLOUD IS THE NEW NORMAL
RESOURCES ARE SCARCE. AS MORE AND MORE ENTERPRISES MAKE THE MOVE TO THE CLOUD, FINDING THE RIGHT PEOPLE TO HELP YOU DRIVE THIS CHANGE IS IMPORTANT, ESPECIALLY IN LIGHT OF SHRINKING BUDGETS AND DECLINING REVENUES
AWS serves hundreds of thousands of customers in more than 190 countries.
Amazon CloudFront and Amazon Route 53 services are offered at AWS Edge Locations