Algorithmic Differentiation, Initial Margin Computation, and Numerical Optimization
This video is the recorded version from the Algorithmic Differentiation stream at QuanTech and includes all the presentation material. There is also a discussion forum included when you purchase this video to continue the debate.
Chair: Uwe Naumann: Professor of Computer Science, RWTH Aachen University, Germany
Quants have been committed to incorporate into pricing more and more XVA adjustments to take into account credit, funding and capital costs exploded after the crisis.
The evolution of this process has led to complex hybrid computation at the level of netting set, and now quants are committed to make computations as efficient as possible. A prominent example is Algorithmic differentiation, that use analytic differentiation to develop a procedure for the computations of large numbers of greeks (sensitivities) at a fraction of the time required by standard finite difference methods.
Keynote speaker: Massimo Morini: Head of Interest Rates, Credit and Inflation Models, Gruppo Intesa Sanpaolo
Blockchain: The New Collateral And Risk Management Paradigm
- Beware of the buzz: what does Blockchain mean?
- Difference between BB (Bitcoin Blockchain) and Permissioned Distributed Ledgers
- Sharing Clearing/Accounting first. Sharing to be fast and automatic
- Financial contracts as pieces of code: an example of a Smart Contract
- What else can be shared and decentralized? Execution, Settlement? Even the Order Book?
- Moving imagination a bit further: Shared but Decentralized Valuation Models
- Difficulties and Opportunities when Quant models are part of contract code
- What room is left for Credit, Funding and Capital problems? A race against Time
- Kill Margin Period of Risk and Redefine Default without risks of market disruption
Panel: The Future of Money & Finance
- Massimo Morini: Head of Interest Rates, Credit and Inflation Models, Gruppo Intesa Sanpaolo
- Michele Curtoni: Strategy Manager, Global Technology Innovation, London Stock Exchange
- Vitalik Buterin: Ethereum founder
- Garrick Hileman: Economic Historian, University of Cambridge and London School of Economics – Founder, MacroDigest.com
- Tomaso Aste: Head Financial Computing & Analytics Group, University College London
- What are the challenges and opportunities facing the finance industry due to the growing influence of new technologies?
- What are the likely disruptive impact of these financial technology?
- Discuss the impact of big data analytics in the financial world
- Is distributed ledgers technology the embryo of a bankless world or the real end of the crisis for banks?
- How banks can win this battle. Winning together or finish first?
The new Algorithmic Pricer Acceleration (APA) and Algorithmic Greeks Acceleration (AGA) methods
Presenter: Ignacio Ruiz: Founder & CEO, MoCaX Intelligence
Efficient XVA Calculations Using The Probability Matrix Method
Presenter: Martin Engblom: Business Development Manager, TriOptima
Vector AAD for Large Portfolios: Applications to FRTB, Dynamic IM, and KVA/MVA*
Presenter: Alexander Sokol: CEO and Head of Quant Research, CompatibL
Compile Time Adjoints via Operator Overloading: Applications to CPU and GPU
Presenter: Jacques Du Toit: Software Developer, NAG
Adjoint Code Design Patterns
Presenter: Uwe Naumann: Professor of Computer Science, RWTH Aachen University, Germany
HPC software using Automatic Differentiation and Multilevel Monte-Carlo methods for financial analytics.
Presenter: Grzegorz Kozikowski: Teaching Assistant, The University of Manchester
Algorithmic Differentiation Panel: Algorithmic Differentiation, Initial Margin Computation, and Numerical Optimization
- Uwe Naumann: Professor of Computer Science, RWTH Aachen University, Germany
- Alexander Sokol: CEO and Head of Quant Research, CompatibL
- Grzegorz Kozikowski: Teaching Assistant, The University of Manchester
- Jacques Du Toit: Software Developer, NAG
- Luca Capriotti: Head QS Global Credit Products EMEA, Credit Suisse & Visiting Professor, University College London (to be confirmed)
- The relative strengths and weaknesses of forward vs. reverse AD mode in practical calculations
- The latest advances in AAD for Monte Carlo and American Monte Carlo
- Practical Implementation of AAD
- “GPU vs. AAD”
- GPUs and AAD are not mutually exclusive – you can do both
- XVA Risk can be done using AAD but regulators and others clearly favour full revaluation for risk and stress testing
- Impact of AAD on Software Engineering practice and strategy
- Globalization of Adjoint Sensitivities through interval arithmetic
- AAD of implicit functions (e.g. non-/linear solvers, optimizers)
- AAD software tools
- Combinatorial problems in AAD