Model Risk Courses
Model Risk in Banking: Goals & Challenges Context
Model risk has become a frontline supervisory priority. Banks are expected not only to develop robust internal models for credit risk (PD, LGD, EAD), market risk, counterparty risk, and capital planning, but also to manage the risks that arise from the models themselves. The rise of AI and ML, coupled with climate and macroeconomic uncertainty, has elevated model risk to a strategic concern for Boards and regulators.
Goals for Banks
-
Robust Governance
-
Implement model risk management frameworks (MRMFs) aligned with ECB, SR 11-7, and EBA standards.
-
Ensure Board and senior management oversight.
-
-
Independent Validation
-
Strengthen validation functions to meet expectations of independence, frequency, and comprehensiveness.
-
Apply quantitative (statistical tests) and qualitative (governance, data quality) validations.
-
-
Integration with Capital Planning
-
Link model risk assessments to ICAAP/ILAAP.
-
Use Margin of Conservatism (MoC) systematically to mitigate deficiencies.
-
-
Adaptation to AI & ML
-
Manage complexity, drift, and explainability risks of ML-driven models.
-
Ensure proportionality: stricter governance for highly complex/dynamic models.
-
-
Regulatory Alignment
-
Comply with ECB Guide to Internal Models (2025) and EBA machine learning guidelines.
-
Demonstrate readiness in on-site model investigations and TRIM-style reviews.
-
Challenges for Banks
-
Complexity of AI/ML Models
-
Black-box risk, lack of transparency, and supervisory demand for interpretability (XAI).
-
-
Data Quality and Availability
-
Need for granular, long-horizon, climate-adjusted, and stress scenario datasets.
-
Managing bias and fairness in AI models.
-
-
Validation Resource Strain
-
Scarcity of skilled validators with expertise in both traditional statistics and ML.
-
High volume of models across risk types.
-
-
Dynamic Model Environments
-
Continuous recalibration and retraining of models vs. static regulatory approval processes.
-
Monitoring drift and material changes.
-
-
Regulatory Pressure
-
Supervisors demanding evidence of robust frameworks: registers, policies, audit trails.
-
Expectation of ICAAP integration and capital add-ons for weak model risk practices.
-
✅ In summary:
Model risk now is about balancing innovation and compliance. The goal is to build a governance and validation ecosystem that is resilient, explainable, and capital-linked. The challenge is doing this in an environment of AI adoption, regulatory scrutiny, and data uncertainty.
Internal Models for Credit Risk: Validation and Model Risk under AI & ML
🎯 Course Objective
The objective of this course is to provide banking and risk management professionals with a comprehensive understanding of internal models for credit risk, their validation frameworks, and the management of model risk in line with ECB guide to internal models July 2025. The course also explores the challenges and opportunities of applying artificial intelligence (AI) and machine learning (ML) within internal models, with a focus on governance, explainability, and regulatory compliance.
Riesgo de Modelo: Cuantificación, Gestión y Machine Learning
en Python y R (Español)
El objetivo del curso es mostrar las mejores prácticas de cuantificación y gestión del riesgo de modelo.
En cuanto a gestión de riesgo de modelo se abordan temas, como la gobernanza, organización, ciclo de vida del modelo, mitigación, model risk appetite y el seguimiento de este riesgo.
Se explican las directivas sobre riesgo de modelo SR 11-7 en EEUU, y la directiva de revisión de modelos internos, TRIM, en la Unión Europea, UE.
Se exponen metodologías avanzadas para medir el riesgo de modelo por la deficiencia de datos, la incertidumbre del parámetro del modelo, el uso inadecuado del modelo, la especificación del modelo, el cambio en la dinámica del entorno financiero y económico y por la implementación incorrecta del modelo.