Forecasting Model Platform for a Large Financial Institution

Oakleaf has delivered services for many years to a large financial institution covering departmental and corporate-wide initiatives, including model applications, databases and platforms for model infrastructure, and integrating analytics tools. Oakleaf has also provided Oakleaf-managed teams and Client-managed supplemental resources. We implemented a human capital strategy tailored to an assessment of skillset gaps and based on continuous conversations in collaboration with hiring managers to ensure that the right talent was available as the project moved through its lifecycle.

This Forecasting Model Platform case study describes our work supporting several corporate-wide initiatives, notably a Big Data platform transformation and analytics engines for new accounting and regulatory requirements such as key earnings metrics and Current Expected Credit Losses (CECL). The challenge of this work included satisfying the needs of multiple economists and risk managers responsible for complex models, and selection and development of the tools and products to provide reports and graphical displays of complex information.

Our Approach

Oakleaf’s engagement with a financial technology department represents years of engagement with their teams supporting their model infrastructure responsibilities, where the company’s principal forecast and historical statistical models are executed, including the runs and reports of outcomes. These reports support analysis and research by sophisticated users without technical programming skills.

One of these applications is the forecasting model platform utilizing SAS, Netezza, Tableau, Java, and Python on an Oracle Database, designed to make forecasting tools easier in the hands of non-technical users. The platform automates the modeling process and incorporates modeling lifecycle management. Users without a knowledge base of SAS or R are able to access its expansive reporting capability through Graphical User Interfaces, generating sophisticated runs with multiple scenarios. Models are profiled as cases, run through this interface and the resulting risk data is displayed in charts and graphs, helping to identify the issues with the run or the data or other issues. The architecture for the forecasting model platform was utilized in other major department applications. Our documentation directly supported this effort.

Oakleaf’s intimate knowledge of the analytic environment, tools, and models of this particular financial institution enabled us to be effective from day one on many new assignments. Our strong command of the datasets and model output analytics on the forecasting model platform project directly contributed to its effectiveness and success, saving the time of many individuals.

Results

Oakleaf’s efforts resulted in faster execution of the models, and therefore the Client’s large volumes of statistical data had understandable and accurate metrics, reports, and graphical dashboards and displays for use by risk management and economists central to their key decision-making.
Countless sets of different-origin data utilized on the platform were cleansed and scrubbed.
The Client was aided in the development and output of reports that compare models, improve sampling, compare outputs, and develop sensitivities. Via the use of these newer processes, the Client is now able to produce reconcilable results through reusable chains of automation runs and an architecture that can be leveraged for other projects.