Tensoring Volatility Calibration: The Power of Chebyshev
22 Oct 2024
About the event
A series of papers have explored the use of Deep Neural Nets to substantially speed up the calibration of pricing models. This paper uses Chebyshev Tensors for the same purpose. In particular; it shows how the computational bottleneck in the calibration of the rough Bergomi volatility model can be alleviated using Chebyhsev Tensors. The calibration speed and accuracy obtained in this paper is comparable to when Deep Neural Nets are used. Building efforts; however; are up to 100 times lower; allowing for much faster pricing model proxy update – a feature of particular importance at times of market distress. This constitutes a further enhancement over the already sizable improvement provided by Deep Neural Nets.
Part of the Fitch Group, Fitch Learning partners with clients to enhance knowledge, skills and conduct. Fitch Learning is a global leader in training with experience of delivering specialised technical training at all levels to the financial community. Fitch Learning partner with clients to elevate knowledge and skills and enhance conduct.
We work with 9 out of 10 of each of the largest Investment Banks, Asset Managers and Global Banks and through state-of-the-art training centres in London, New York, Hong Kong, Singapore and Dubai, and our leading distance learning portals, we train more than 20,000 delegates each year.
