I am delighted to present the February 2019 issue of The Journal of Computational Finance.
In our first paper, “A pairwise local correlation model”, Frank Koster and Daniel Oeltz introduce a novel local correlation model in which, using a particle method, they calibrate a generic function of pairs of asset values to the local variance of an index. They discuss in detail the use of the model for the pricing of further derivative instruments.
Leading on from their seminal 2007 Journal of Computational Finance paper on the numerical approximation of optimal stochastic control problems in finance, Peter A. Forsyth and George Labahn present “ε-monotone Fourier methods for optimal stochastic control in finance”, this issue’s second paper. In it, the authors combine the efficiency of Fourier methods with the no-arbitrage properties of monotone projection schemes to overcome the classical shortcomings that these methods have individually.
In “Dilated convolutional neural networks for time series forecasting”, another story of the success of machine-learning approaches published in this journal, Anastasia Borovykh, Sander Bohte and Cornelis W. Oosterlee conduct extensive and rigorous tests on the performance of a convolutional neural network architecture for multivariate time series forecasting, benchmarking against standard autoregressive models, among others.
Finally, Christian P. Fries demonstrates the efficiency of automatic differentiation in backward mode for the computation of forward sensitivities in “Fast stochastic forward sensitivities in Monte Carlo simulations using stochastic automatic differentiation (with applications to initial margin valuation adjustments)”. These sensitivities are essential in, for instance, valuation adjustments.
Let me take this opportunity to announce a special issue of The Journal of Computational Finance to mark the occasion of the International Conference on Computational Finance, which will be held in the beautiful Galician city of A Corun˜ a from July 9 to July 12, 2019. We will be communicating details with regard to submission and the Best Paper Prize shortly. We are honored that a number of Journal of Computational Finance editors are instrumentally involved in the conference, either on the scientific committee or as plenary speakers. We are looking forward to seeing many of you there, as we are always eager to hear your thoughts on the direction in which this field is going and the role our journal has to play in it.
In the meantime, I hope you enjoy the present issue.
University of Oxford
In this paper, the authors develop a new local correlation model that uses a generic function 'g' to describe the correlation between all asset–asset pairs for a basket of underlyings.
In this paper, the authors give a preprocessing step for Fourier methods that involves projecting the Green’s function onto the set of linear basis functions.
In this paper, the authors present a method for conditional time series forecasting based on an adaptation of the recent deep convolutional WaveNet architecture.
Fast stochastic forward sensitivities in Monte Carlo simulations using stochastic automatic differentiation (with applications to initial margin valuation adjustments)
In this paper, the author applies stochastic (backward) automatic differentiation to calculate stochastic forward sensitivities.