Advanced Model Risk

MODEL VOLATILITY – PANEL DISCUSSION

10:00 Reviewing model reaction to COVID-19 and assessing the impact of volatility in 2021

• How models adapted to changing environment in 2020
• Impact of changes to market behaviors and macroeconomic variables
• Managing capital models
• Maintaining models as variables continue to change

REGULATION

10:40 Reviewing current and future changes to regulation and applicability as models and applications continue to evolve

• Changes to LIBOR models
• SR 11-7 regulatory guidance
• What regulators are looking for when conducting examinations
• Understanding model usage and inventory
• Guidance and management for international institutions
  • International, regional and ring fenced entities
• Standards accepted for emerging methods and techniques
• Assessing new regulatory advances in 2021

11:10 Networking break

AUTOMATION

11:30 Reviewing potential uses of automation across institutions and the role of model risk teams

• What is it and what is being done?
• Automating validation tasks and reports
• Cutting down time and resource heavy tasks
• Impact to industry and team structure
  • Monitoring automation
• Drivers for model risk team sizes
• Skillsets and resources required ahead

ETHICS AND BIAS – PANEL DISCUSSION

12:00 Ethics and machine learning: Removing and continuously monitoring for bias in applications

• Regulations across institutions and industries
  • BigTech regulation
• Protecting customers from bias
• Model explainability to understand outputs
• Reporting results to regulators

12:40 Governance and validation of machine learning models to better monitor and understand outputs

• Implementing a framework for management of AI and ML
• Validation challenges with complex models
  • Neural networks and self-reinforcement methodologies
• Monitoring and surveillance uses
• Use of vendor models for AI/ML capabilities vs. building internally
• Explainability and interpretability of AI and ML models

1:10 Closing remarks and end of day one

NON-QUANTITATIVE INPUTS

10:00 Interpretation and treatment of untraditional models using non-quantitative inputs

• Managing technology including voice recognition
• Managing qualitative data inputs
• Modifying model governance requirements
• Enhancing risk management standards on non-quantitative inputs
• New standards to account for changes in technology
• Regulatory guidance to account for non-quantitative inputs

INVENTORY – PANEL DISCUSSION

10:30 Ensuring completeness and accuracy of model inventory to monitor uses across all areas

• Identifying what models are in use in different areas
• Cross section of model population for specific exercises
• Ensuring model owners understand guidance and frameworks
• Understanding all uses for a specific model
  • Validation requirements when using a model for purposes it was not built for

11:10 Networking break

MODEL BOUNDARY

11:30 Defining boundaries for effective oversight and governance of inventory

• FinTech and financial institution barriers
• Interpretation and potential for bias
• Maintaining monotonistic constraints – criteria to impose on models
• Understanding factors impacting models
• Providing predictive power to models and explainability

MODEL VALIDATION

12:00 Updating validation approaches and techniques with continued advances in model inventory and definition

• Expanding model risk to incorporate AI/ML applications
• Integrating control framework approach to non-models
• Improving efficiencies through AI and machine learning adoption
• Monitoring the pace of change to benefit customers
• Strengthening risk and accelerating adoption
• Balancing quantitative and risk thinking in a model risk team
• Streamlining model risk practices to manage risks with emerging practices and applications
• Model risk as a key contributor to risk management

DEEP LEARNING

12:30 Leveraging deep learning capabilities to better serve customers and make informed credit decisioning

• Credit risk implications with more accurate decisioning
  • Testing effectiveness of outputs and explainability
• Uses within fraud and improving stress testing
• Privacy, fair lending and bias risks
• Bridging the skill gap between AI and finance
• Understanding societal impact of deep learning applications at development
• Use of smart and alternative data
  • Gaining broader insights for increasingly complex models

1:00 Closing remarks and end of Advanced Model Risk Forum 2021

ADVANCED MODEL RISK IS A FREE TO ATTEND EVENT…

IF YOU ARE REPRESENTING A SERVICE PROVIDER, VENDOR OR CONSULTANT…

You are able to register for a nominal fee of $599. Online registration will be made available shortly.
All registrations are manually reviewed and approved before any confirmation emails are sent.
If you would like to register or have any questions please email alex@cefpro.com
CeFPro has the final word on all registrants