When I wrote my letter for the previous issue of The Journal of Computational Finance, I hoped that by June most of us would not still be forced to work from home. Sadly we still are, but one positive result of this crisis has been the emergence of numerous exciting online seminars and lectures that we can attend without get-ting on a plane. For instance, the SIAM Activity Group on Financial Mathematics and Engineering now orchestrates a virtual seminar series, interleaved with one run by the Bachelier Finance Society. The upcoming workshop “Machine Learning in Quantitative Finance and Risk Management”, which will be held on July 2 and has been co-organized by Kees Oosterlee, a previous editor-in-chief of The Journal of Computational Finance, may also be of interest to you.
In the fast-evolving field of machine learning in finance, the wealth of papers that has recently been written makes it hard to maintain an overview of all the latest developments. In this issue’s first paper, “Neural networks for option pricing and hedging: a literature review”, Johannes Ruf and Weiguan Wang help us out by providing a comprehensive review of the use of neural networks in derivatives pricing and hedging. They also offer their own recommendations for the application of these new techniques from their study of the literature.
Following this, our second paper is “Gaussian process regression for derivative portfolio modeling and application to credit valuation adjustment computations”. Here, Stephane Crepey and Matthew F. Dixon introduce a Gaussian process meta- model for derivative portfolios in terms of market factors, giving applications to credit valuation adjustment computations where the whole mark-to-market cube can be efficiently approximated.
In the third paper in the issue, “High-order approximations to call option prices in the Heston model”, Archil Gulisashvili, Marc Lagunas-Merino, Raul Merino and Josep Vives derive accurate approximation formulas to the Heston model by expansion in the volatility-of-volatility parameter. In doing so, they demonstrate that the approximation is both more accurate than the others that are available and significantly faster to obtain than a semianalytical method based on the fast Fourier transform.
Finally, Christian P. Fries introduces an extension to classical term-structure models that allows for dynamic tenors in “Dynamic refinement of the term structure: time-homogeneous term structure modeling”. The author also derives no-arbitrage conditions and provides a practically efficient implementation in his paper.
I hope you are keeping safe and well and will find inspiration in these papers.
University of Oxford
This paper provides a comprehensive review of the field of neural networks, comparing articles in terms of input features, output variables, benchmark models, performance measures, data partition methods and underlying assets. Related work and…
Gaussian process regression for derivative portfolio modeling and application to credit valuation adjustment computations
The authors present a multi-Gaussian process regression approach, which is well suited for the over-the-counter derivative portfolio valuation involved in credit valuation adjustment (CVA) computation.
In the present paper, a decomposition formula for the call price due to Alòs is transformed into a Taylor-type formula containing an infinite series with stochastic terms. The new decomposition may be considered as an alternative to the decomposition of…
The author considers a classical term structure model framework, ie, a Heath–Jarrow–Morton framework, on a time-discrete tenor, such as the London Interbank Offered Rate market model, using a sequence of tenor discretizations, where the tenors are valid…