Programme Overview
AI and quantitative methods for complex, extreme and emerging risk
Financial and insurance institutions increasingly face risks that are rare, interconnected, fast-moving or poorly represented by historical averages. Climate and environmental shocks, cyber incidents, liquidity stress, contagion and extreme losses challenge both traditional models and purely data-driven solutions.
At the same time, modern statistical learning, machine learning, deep learning and generative AI provide powerful ways to work with complex data, detect structure, construct scenarios and monitor changing behaviour. These methods create value only when they are grounded in an appropriate risk framework, validated carefully and used with a clear understanding of their limitations.
This CAS brings those elements together. It develops rigorous foundations for quantitative risk analysis and connects them to contemporary AI methods, realistic applications and responsible model governance.
Workload. The programme offers an overall of 90 classroom hours, plus guided homework, assessments and approximately 90-hour capstone. Overall, this corresponds to 15 ECTS , that is, approximately 450 hours total workload.
Python preparation. Upon admission, you may be required to take a Python preparation pre-course (videos). The videos are made available without additional cost, but they require and additional worload (depending on the candidate's previous programming preparation) not included in the calculation of the workload above.
Curriculum 2027
Module 1 - Quantitative Foundations for Risk Analysis
Module overview. The module combines mathematical tools for risk analysis - extreme value theory, large deviations, dependence models and point processes - with the foundations of quantitative risk management, including risk concepts, pricing, aggregation, diversification, capital, capital allocation and internal models.
|
Credits |
4 ECTS |
|
Teaching allocation |
5.5 days March 2027: 5, 6, 12, 13, 19, 20 (half day) |
|
Module lecturers |
Prof. Ashkan Nikeghbali; Dr. Michel Dacorogna |
Module 2 - AI for Risk Modelling, Validation and Monitoring
Module overview. The module introduces statistical and model-based learning, supervised and unsupervised machine learning, deep learning and black-box decision search.
|
Credits |
3 ECTS |
|
Teaching allocation |
5 days April 2027: 16, 17 (half day), 23, 24 May 2027: 21 (half day), 22 |
|
Module lecturers |
Prof. Reinhard Furrer; Prof. Anastasiia Koloskova; Prof. Nicola Serra; Prof. Jan Wegner |
Module 3 - Explainable and Ethical AI
Module Overview. This module gives insurance and finance professionals the concepts and practical tools required to explain, audit and improve machine-learning models used in risk analysis. It covers interpretable models, black-box explanations and responsible-AI practices built on fairness, ethics, accountability, explainability, privacy, security and governance. It also situates these tools within current industry practice in explainable AI. Each concept is illustrated with insurance cases such as pricing, scoring and underwriting and connected to the regulatory frameworks named in the source material: the EU AI Act, the NIST AI Risk Management Framework and US state insurance bulletins on algorithmic bias.|
Credits |
2 ECTS |
|
Teaching allocation |
1 day May 2027: 29 |
|
Module lecturer |
Dr. François Hu |
Module 4 - Applications Studios: Emerging Risks
Module overview. The module frames emerging risks, considers model strategy and risk mitigation, and develops applications in cyber risk, liquidity and systemic risk, and climate/environmental shocks.
|
Credits |
3 ECTS |
|
Teaching allocation |
3.5 days June 2027: 4, 5, 18, 19 |
|
Principal application areas |
Cyber risk; liquidity and systemic risk; climate risk |
|
Module lecturers |
Prof. Alexandre Bovet; Prof. Caroline Hillairet; Prof. Reinhard Furrer; Prof. Jan Wegner |
Module 5 - Capstone Project
The Capstone Project (3 ECTS) brings together the quantitative, data-driven and AI methods developed throughout the programme. Participants work on a concrete risk-analysis problem, applying appropriate modelling and computational tools to real-world or realistic data. Projects may address topics from finance, insurance or emerging risks and are developed with academic guidance.
The project culminates in a written report and presentation, demonstrating the ability to formulate a risk problem, select and critically assess suitable methods, interpret results, and communicate their implications for risk management and decision-making.
Example Capstone Projects
Insurance & actuarial risk
Build and validate an AI-enhanced insurance pricing model, with particular attention to tail risk, calibration and explainability.
Systemic & financial risk
Analyse a financial network and simulate how liquidity or credit shocks propagate across interconnected institutions.
Climate & cyber risk
Combine data-driven methods with extreme-risk modelling to quantify physical climate losses or cyber accumulation scenarios.
AI validation & emerging risk
Develop a framework for monitoring an AI risk model under distribution shift, or use generative AI to detect emerging risks from unstructured information.