Description: In order for an organization to benefit from artificial intelligence/machine learning (AI/ML) models, it requires the involvement of different business areas. During this webcast, professionals from data analytics, audit and risk areas will address challenges and provide solutions for working together to develop, implement, use and risk manage AI/ML tools. Throughout the discussion, the presenters will discuss: • Defining and exploring the challenges and objectives between data scientists, domain experts and control functions • Risk management – model risks and other risks • End-to-end model validation • Model performance vs. interpretability/understandability trade-offs • What to audit – auditor’s challenge and solutions

Description: Modeling Approaches and Challenges focuses on approaches to real world simulation and the challenges and considerations in real world simulation. Topics include: A frequency domain approach to build realistic multi-horizon scenarios. (Ortec approach) Cascade models structure linking models for major asset classes and economic variables. (Moody’s approach) Overview of models, calibrations & approaches within the context of different use cases. Deterministic & stochastic approaches. Importance of model re-calibration flexibility. We will also discuss specific insights into considerations and challenges in real world simulation, which include topics like: Setting capital market assumptions and/or calibration targets. Assumptions about correlation and causality. Calibrating the model to capital markets assumptions. Validating the model against capital market assumptions. Role of history, market data and stylized facts. Importance of back testing. Calibrating the model to multiple investment horizons. Inclusion of alterative asset classes such as private equity, infrastructure, and direct real estate. Associated issues of limited data, illiquidity and greater uncertainty about calibration targets.