Abstract:Stochastic differential equation (SDE) models are the foundation for pricing and hedging financial derivatives. The drift and volatility functions in SDE models are typically chosen to be algebraic functions with a small number (less than 5) parameters which can be calibrated to market data. A more flexible approach is to use neural networks to model the drift and volatility functions, which provides more degrees-of-freedom to match observed market data. Training of models requires optimizing over an SDE, which is computationally challenging. For European options, we develop a fast stochastic gradient descent (SGD) algorithm for training the neural network-SDE model. Our SGD algorithm uses two independent SDE paths to obtain an unbiased estimate of the direction of steepest descent. For American options, we optimize over the corresponding Kolmogorov partial differential equation (PDE). The neural network appears as coefficient functions in the PDE. Models are trained on large datasets (many contracts), requiring either large simulations (many Monte Carlo samples for the stock price paths) or large numbers of PDEs (a PDE must be solved for each contract). Numerical results are presented for real market data including S&P 500 index options, S&P 100 index options, and single-stock American options. The neural-network-based SDE models are compared against the Black-Scholes model, the Dupire's local volatility model, and the Heston model. Models are evaluated in terms of how accurate they are at pricing out-of-sample financial derivatives, which is a core task in derivative pricing at financial institutions.

Comparison Tests of Variable-Stepsize Algorithms for Stochastic Ordinary Differential Equations of Finance

Variable stepsize Runge-Kutta methods for stochastic wave equations

Stochastic PDEs and Quantitative Finance: The Black-Scholes-Merton Model of Options Pricing and Riskless Trading

Parameter Estimation for Multivariate Nonlinear Stochastic Differential Equation Models: A Comparison Study

Option Valuation using Fourier Space Time Stepping

Machine Learning Methods for Pricing Financial Derivatives

Stochastic Differential Equations for Modeling First Order Optimization Methods

A brief review on stability investigations of numerical methods for systems of stochastic differential equations

On Pricing Options under Two Stochastic Volatility Processes

An efficient numerical approach for solving three‐dimensional Black‐Scholes equation with stochastic volatility

A Comparison of Reinforcement Learning and Deep Trajectory Based Stochastic Control Agents for Stepwise Mean-Variance Hedging

Numerically Unveiling Hidden Chaotic Dynamics in Nonlinear Differential Equations with Riemann-Liouville, Caputo-Fabrizio, and Atangana-Baleanu Fractional Derivatives

Positivity-preserving truncated Euler and Milstein methods for financial SDEs with super-linear coefficients

Differential Equations for Modeling Asynchronous Algorithms

Exact retrospective Monte Carlo computation of arithmetic average Asian options

$ε$-Monotone Fourier Methods for Optimal Stochastic Control in Finance

Strongly asymptotically optimal schemes for the strong approximation of stochastic differential equations with respect to the supremum error

Probabilistic evolution of the error of numerical method for linear stochastic differential equation

Stochastic Modified Equations and Dynamics of Stochastic Gradient Algorithms I: Mathematical Foundations

A new weak approximation scheme of stochastic differential equations and the Runge-Kutta method

Performance Evaluation of Single-step Explicit Exponential Integration Methods on Stiff Ordinary Differential Equations