Paper: arXiv 2403.14231
Authors: Sébastien Bossu, Stéphane Crépey, Hoang-Dung Nguyen
Abstract
We propose a distributional formulation of the spanning problem of a multi-asset payoff by vanilla basket options. This problem is shown to have a unique solution if and only if the payoff function is even and absolutely homogeneous, and we establish a Fourier-based formula to calculate the solution. Financial payoffs are typically piecewise linear, resulting in a solution that may be derived explicitly, yet may also be hard to numerically exploit. One-hidden-layer feedforward neural networks instead provide a natural and efficient numerical alternative for discrete spanning. We test this approach for a selection of archetypal payoffs and obtain better hedging results with vanilla basket options compared to industry-favored approaches based on single-asset vanilla hedges.
Complexity vs Empirical Score
- Math Complexity: 8.0/10
- Empirical Rigor: 5.5/10
- Quadrant: Holy Grail — high math complexity, high empirical rigor
Why this score: The paper introduces an innovative theoretical framework using Fourier transforms and neural network activations (ReLUs) to solve the multi-asset payoff spanning problem. While the mathematical derivations are rigorous and novel, the empirical section focuses on standard archetypal payoffs rather than extensive market backtests.
Research Flowchart
flowchart TD
A["Research Goal: Span Multi-Asset Payoffs<br/>using Vanilla Basket Options"] --> B["Distributional Formulation &<br/>Fourier-Based Solution"]
B --> C{"Conditions Met?<br/>(Even & Absolutely Homogeneous)"}
C -- "Theoretical Exact Solution" --> D["Piecewise Linear Payoffs<br/>(Hard to exploit numerically)"]
D --> E["1-Hidden-Layer Neural Network<br/>(Efficient Discrete Spanning)"]
F["Data Inputs: Archetypal Financial<br/>Multi-Asset Payoffs"] --> E
E --> G["Key Outcome: Superior Hedging Performance<br/>vs. Single-Asset Industry Approaches"]
Cited References (69)
- The loss surfaces of multilayer networks (2015)
- Universal approximation with deep narrow networks (2020)
- Approximation by superpositions of a sigmoidal function (1989)
- Approximation capabilities of multilayer feedforward networks (1991)
- Explainable neural network for pricing and universal static hedging of contingent claims (2022)
- An overview of artificial neural networks for mathematicians (2018)
- Quadrature on a spherical surface (2015)
- How do infinite width bounded norm networks look in function space? (2019)
and 61 more
- Spherical quadrature formulas with equally spaced nodes on latitudinal circles (2009)
- A function space view of bounded norm infinite width R elu nets: The multivariate case (2020)
- Static Replication of European Multi-Asset Options with Homogeneous Payoff (2022)
- Analytic Approximations for Multi-Asset Option Pricing (2012)
- Prices of state-contingent claims implicit in option prices (1978)
- Towards a theory of volatility trading (1998)
- Repricing the Cross Smile: An Analytic Joint Density (2011)
- Multi-asset Stochastic Local Variance Contracts (2011)
- A new representation of the risk-neutral distribution and its applications (2022)
- Static replication of E uropean standard dispersion options (2022)
- Regularization for deep learning: A taxonomy (2017)
- Batch normalization: Accelerating deep network training by reducing internal covariate shift (2015)
- Extreme learning machine: a new learning scheme of feedforward neural networks (2004)
- Deep residual learning for image recognition (2016)
- Approximation theory of the MLP model in neural networks (1999)
- Optimization of conditional value-at-risk (2000)
- Adam: A Method for Stochastic Optimization (2015) — link
- Extreme learning machine: theory and applications (2006)
- Equity Derivatives Corporate and Institutional Applications (2017)
- Advanced Equity Derivatives (2014)
- Numerically stable generation of correlation matrices and their factors (2000)
- Alternatives to Deep Neural Networks in Finance (2022)
- Multiasset Derivatives and Joint Distributions of Asset Prices (2014)
- Optimal static-dynamic hedges for exotic options under convex risk measures (2009)
- Non-parametric Pricing and Hedging of Exotic Derivatives (2020) — link
- Neural joint S&P 500/VIX smile calibration (2022)
- Matrix Computations (2013)
- A First Course in Fourier Analysis (2008)
- A model-free approach to continuous-time finance (2023) — link
- Generalized Functions: Theory and Applications (2004)
- Hilbert Transforms (2009)
- Hilbert transforms in signal processing (1996)
- Impulse functions over curves and surfaces and their applications to diffraction (2006)
- The formula of Faa di Bruno (1980)
- Hypersingular integrals: how smooth must the density be? (1996)
- Singular Integral Equations (2000)
- The Mathematics of Computerized Tomography (2001)
- The Radon Transform and Some of Its Applications (1983)
- Über die Bestimmung von Funktionen durch ihre Integralwerte längs gewisser Mannigfaltigkeiten (1917)
- Complex variables and applications (2009)
- Cauchy principal value contour integral with applications (2017)
- Calculus: early transcendentals (2020)
- Introduction to Radon Transforms (2015)
- The Radon Transform and Local Tomography (1996)
- Probability density estimation using delta sequences (1979)
- Principles of Mathematical Analysis (1976)
- Functional Analysis (1980)
- L$^p$ boundedness of the H ilbert transform (2009) — link
- Tempered distributional Fourier sine (cosine) transform (2006)
- Meijer G-functions: a gentle introduction (2013)
- MeijerG (1996)
- On the p-norm of the truncated H ilbert transform (1988)
- Introduction to the Theory of Distributions (1998)
- Fourier transforms of odd and even tempered distributions (1974)
- Decay of the Fourier Transform (2014)
- Basic Real Analysis (2007)
- On certain representations of pricing functionals (2024)
- A F ourier Transform Method for Spread Option Pricing (2010) — link
- Learning theory from first principles (2024)
- Foundations of machine learning (2018)
- Optimal positioning in derivative securities in incomplete markets — → 2403.00139
1 of 69 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2518 linked papers.