University of Oxford
It is my pleasure to bring to you the last of 2023’s issues of The Journal of Computational Finance.
In our first paper, “Neural variance reduction for stochastic differential equations”, Piers Hinds and Michael Tretyakov use a deep learning strategy to construct approximately optimal control variates of estimators. By using a parameterization of conditional expectations and leveraging a state-of-the-art optimization methodology, they achieve variance reductions (and hence computational efficiency gains) of several orders of magnitude.
>In “Optimal damping with a hierarchical adaptive quadrature for efficient Fourier pricing of multi-asset options in Lévy models”, the issue’s second paper, Christian Bayer, Chiheb Ben Hammouda, Antonis Papapantoleon, Michael Samet and Raúl Tempone consider, instead of simulation, a semi-analytic strategy that is applicable to models with a known characteristic function. They provide a comprehensive solution to tackle the typical challenges of these methods, chief among them the weak regularity of the integrand and the curse of dimensionality.
In the third paper in this issue, Marilena Furno presents “Extremiles, quantiles and expectiles in the tails”, which analyzes these different regression estimators for the tail risk and provides valuable insights into their behavior for different distributions.
In “Evaluating credit valuation adjustment with wrong-way risk for Bermudan options”, our fourth and final paper, Bing Dong, Wei Xu and Guangguang Wang study the application of tree-type Markov chain approximations to stochastic models for counterparty credit risk computations in the context of early exercise and wrongway risk.
Finally, I would like to take this opportunity to offer my congratulations to the journal’s Associate Editor, Artur Sepp, who has been named Risk Awards 2024 Quant of the Year. Artur has an impressive track record of instrumental contributions to the field, and he was specifically recognized for his recent work with Parviz Rakhmonov on “A robust stochastic volatility model for interest rates” (Risk, September 2023, pp. 1–6), which introduces a versatile model for positive and negative volatility correlations that captures the yield curve accurately and calibrates efficiently to swaptions. Huge congratulations, Artur!
I wish everyone a successful end of the year and hope that you all find some time for relaxing and recharging.
This paper proposes the use of neural stochastic differential equations as a means to learn approximately optimal control variates, reducing variance as trajectories of the SDEs are simulated.
Optimal damping with a hierarchical adaptive quadrature for efficient Fourier pricing of multi-asset options in Lévy models
The authors put forward a a method for pricing European multi-asset options intended to address challenges related to the choice of damping parameters and the treatment of high dimensionality when designing methods for Fourier pricing options.
The author investigates quantiles, expectiles and extremiles in tail estimators for linear regression.
The authors propose a method for credit valuation adjustment evaluation that avoids the need for simulation while maintaining efficiency and accuracy.