AAD for Derivatives Risk Management Workshop
Led by Antoine Savine, Quantitative Researcher, Danske Bank
Monday 11 May 2020
Your 2020 workshop leader
Antoine Savine - Quantitative Research - Danske Bank
Antoine Savine is a mathematician, academic and a leading derivatives practitioner with Superfly Analytics at Danske Bank, winner of the Risk In-House System of the Year 2015 award and the Excellence in Risk Management and Modelling RiskMinds 2019 award.
Antoine has held multiple leading positions in quantitative finance, including Global Head of Research at BNP-Paribas 1999-2012.
Danske Bank’s multi-award winning risk management platform combines a number of cutting-edge technologies, like Algorithmic Adjoint Differentiation (AAD), model hierarchies, cash-flow scripting, parallel Monte-Carlo or Deep Learning. Antoine’s annual talks with QuantMinds and RiskMinds explain these technologies and their implementation in derivatives risk management.
Antoine is the author of the Modern Computational Finance book, published with Wiley, which describes and explains these technologies in deep detail with complete, professional C++ code.
Your 2020 workshop highlights
Arguably the strongest addition to numerical finance of the past decade, Algorithmic Adjoint Differentiation (AAD) is the technology implemented in modern financial software to produce thousands of accurate risk sensitivities, within seconds, on light hardware.
AAD recently became a centerpiece of modern financial systems and a key skill for all quantitative analysts, developers, risk professionals or anyone involved with derivatives. It is increasingly taught in Masters and PhD programs in finance.
Danske Bank's wide scale implementation of AAD in its production and regulatory systems won the In-House System of the Year 2015 Risk award.
Deep Learning and Derivatives Finance
- Brief introduction to neural networks and deep learning
- Application to Derivatives risk management
Implementation in Python and TensorFlow
- Computation graphs and back-propagation, in deep detail
- Adjoint differentiation as generalized back-propagation
- Overview of applications in machine learning and finance
AAD in C++*
- Development of a simple AAD framework, from scratch, in standard C++
- Implementation of recording with operator overloading
- Implementation of adjoint propagation
- Simple applications
*Sessions 3 and 4 discuss the practical development and application of AAD in C++. We produce concise, simple code in standard C++, developing from scratch a simplified, yet fully functional implementation AAD and its application to Monte-Carlo risk sensitivities
Application to Derivatives Risk Management
- Application to non-trivial Derivatives risk reports, like exotics and regulations with Monte-Carlo
- Example from scratch in standard C++: vega reports in Dupire's model
- Extensions and conclusion