Strategy

Capital Stress Testing (CCAR, HCR)

Skills: Python (Pandas, Sklearn, Matplot), Statistics, PowerPoint, Presenting, Governance, Leadership, CECL, Finance, Economic Forecasting,

Stress Testing Models and Methodology is critical for the process of strategic decisioning. With forethought the setup and implementation of a good Capital Stress Testing process supports CECL, Budgeting, and regulatory Stress Testing.

At DCU I completed the first ever Capital Stress Testing (CPST), two years ahead of schedule. We strategically decided to do a quick, simple rollout that would enable us to begin building 'the pipes', conversations, and reporting to support a massive leap in our capabilities.

The final vision is a single system bridging many financial forecasting needs with the final product serving in the CPST regulatory requirements. Along the way it will be able to answer a variety of strategic decisions and support our core businesses. We counting on positive feedback loops between governance, review, and reporting while reducing costs.

Source: Moodys.com

Statistics, Machine Learning, and Data Science

Developed machine learning (logistic) models to forecast defaults across all consumer lending products.

The presentation at left, is based on governance I developed for the models I created. Models are loan level using millions of loans over several years time.

These models are incorporated most commonly in risk awareness. However their largest impacts are for strategic decisioning such as cutting/extending lines for credit and in marketing initiatives. At DCU we have been using these for several years for Credit Card marketing promotions. Additionally we did testing to support the credit risk portion of profitability and pricing via AIRB.

Skills: Python (Pandas, Sklearn, Matplot), SQL, Statistics, Machine Learning, PowerPoint, Presenting, Governance, Leadership

Programming

Programming is where I gain efficiencies over time. I plan my various projects in a coordinated way to gain efficiencies in quality, accuracy, and automation. By architecting modules, I am able to easily borrow code between projects that can also be used to support other project members. Critically with the rise of Copilot's, my productivity soars as I can more closely consider program architecture, outputs, and demonstrations.

On a daily basis I programming in Python, which I have been utilizing since 2015. My roots for programming extend back to Excel (Power User), R* and SAS, but I am primarily in Python these days.

I spend a lot of work hours looking at risks of individual companies and individuals. In my personal time I am starting to build models looking at the health of a network or environment. This would be useful for understanding inter-relationship dependencies especially where liquidity and credit is a concern.

Mentoring

Net Present Value is the classic 'best' decisioning tool for evaluating and comparing projects. I demonstrated to a group of Data Scientists th calculation, evaluation, and decisioning process. The examples I used were complete with different economic scenarios (recession vs non), different revenue streams, and potential for optional offshoot pojects.

To round out this demonstration, I walked through ROI and Payback Period and how they might return incorrect results.

Presenting: Project Decisioning

Team Work to Dream Work

What everyone needs varies. Between coaching tennis and managing, I find it's best to try and meet people where they are. At the beginning that can be things like teaching them to organize their thoughts for a productive mentoring session. As they mature it's increasingly letting them run the show on various conversations and projects.

As they grow and learn, my trust grows with them where I illicit their feedback on how my projects could be better.

Wherever we are in the process, it's important to build the trust and communication together.