Complex environments require complex systems to model. In some cases machine learning techniques can be applied successfully, in other situations, we may need to address them individually. Join experts from QuantMinds International 2022 and discover new techniques in risk modelling.
Projecting exposures and margin: Getting risk models and pricing models to play nice
Andrew McClelland, Director, Quantitative Research at Numerix, presents his work on projecting off a dedicated risk model (discrete-time PCA etc.), the complex translation between the risk model and the pricing model on a per-scenario basis. Learn about:
- The problem of projecting exposures and margin requirements off a pricing model (arbitrage free etc.) is well understood
- We have a range of numerical techniques (e.g. LSMC) to do this efficiently when closed-form representations are not available
- But what if we are projecting off a dedicated risk model (discrete-time PCA etc.)?
- Translating between the risk model and the pricing model on a per-scenario basis is complex, and it may involve calibration
- We investigate this problem and identify a natural classification according to the type of risk model employed
- Some classes are amenable to simple extensions of standard techniques while others will require more advanced approaches
Practical implementation of machine learning techniques for risk and pricing
Enjoy this session from QuantMinds International, in which Christian Kappen, Manager at d-fine, shares his work on practical implementation of machine learning techniques for risk and pricing. Learn about:
- Machine learning in finance – A retrospection
- Machine learning in risk and pricing – The supervisory perspective
- Machine learning in pillar II market risk – A pre-study at NRW.BANK
- Description of the use case
- Fast pricing of Bermudians in VaR: Machine learning approaches
- Model lifecycle management
- Technical implementations
- Results
- Outlook
