PV and XVA Greeks for Callable Exotics: A New Approach to Algorithmic Differentiation
Pricing and risk management of OTC Derivatives became immensely more complex with the introduction of XVAs. Among the key challenges introduced was further complication of the calculation of risk sensitivities, PV and Greeks. An already onerous task before XVAs, the traditional approach requires numerous time intensive calculations run in multiple steps making it both complex and time consuming. A presence of semi-exotic and exotic instruments in the portfolio makes the problem much more difficult due to numerically involved methods, like the American Monte Carlo (regression).
That lengthy computation time can have great opportunity cost and risk exposure to banks, not to mention high compute demand, which lead to the mandate for greater speed and performance through numerical optimizations like adjoint differentiation (AD).
AD simplifies the process by calculating the sensitivities simultaneously using a single pricing calculation, and results in substantial performance improvement. However, AD coding is technically difficult and at times leads to the rewriting of entire pricing libraries. Additionally, in traditional AD the calculation of the Greeks takes place after the main pricing calculation and therefore it is necessary to record information generated during pricing to use later on. Moreover, the exposure calculation of semi-exotic and exotic instruments requires regression techniques which lead to extra complications of the AD application resulting in substantial memory consumption.
On June 7th featured speaker Alexandre Antonov, SVP of Quantitative Research at Numerix, shared his latest research on a new approach to XVA Greeks for general instruments called the Backward Differentiation (BD) technique. It is performed at the same time as the pricing which leads to a smaller memory consumption and makes the implementation much easier compared with to the standard AD.
Dr. Antonov covered the following topics:
-
Algorithmic Differentiation for PV Greeks in the nutshell
-
Algorithmic Differentiation for exotic instruments
-
Monte Carlo regression: Adjoint Differentiation vs Backward Differentiation
-
Discussion on necessity of the regression differentiation
-
Numerical experiments: XVA for a Bermudan swaption
Attendance is complimentary, Registration is required. Space is limited, reserve your seat today!
Featured Speaker:
Dr. Alexandre Antonov, Senior Vice President of Quantitative Research, Numerix
Dr. Antonov was Risk Magazine’s 2016 Quant of the Year. He first began studying Physics and Mathematics at the Moscow Institute of Physics and Technology, followed by the Landau Institute for Theoretical Physics (PhD in 1997) and The Laboratory of Theoretical and High Energy Physics at the University of Paris VI.
In March 1998, Alexandre joined Numerix based in Paris taking on a number of positions during his tenure with the company including Quantitative Analyst, Senior Quantitative Analyst, Vice President and Senior Vice President. During the course of his career Dr. Antonov has had over twenty articles published in industry journals, including six that have appeared on the pages of Risk Magazine.
Moderator: Cristin Riffle-Lash, Vice President of Marketing, Numerix
Mrs. Riffle-Lash leads the company's thought leadership program. Through the program, she works closely with global Numerix experts, clients and partners to breakdown industry challenges, elevate strategies, disseminate industry-leading research, and shine a light on break-through solutions and innovations. She joined Numerix in 2012, and as VP of integrated marketing directs digital marketing, content marketing and relationship marketing across the firm.
Prior to joining Numerix, she worked with finance and capital markets luminaries heading up marketing for the Wiley Finance series; GARP, PRMIA and CFA Publishing relationships; and the Wiley Trading Series. Mrs. Riffle-Lash has more than 10 years of experience across financial services and B2B marketing, including roles at Institutional Investor and Springer Verlag. She has a B.S. in Magazine Journalism and a B.S. in Psychology from the University of Florida.
