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.

Option pricing under path-dependent stock models

Pricing Exotic Options in a Path Integral Approach

Analytical and Numerical Approaches to Pricing the Path-Dependent Options with Stochastic Volatility

Option pricing: A yet simpler approach

Option Pricing with Stochastic Volatility, Equity Premium, and Interest Rates

Monte Carlo Methods and Path-Generation techniques for Pricing Multi-asset Path-dependent Options

Path Integral Method for Pricing Proportional Step Double-Barrier Option with Time Dependent Parameters

Pricing Options under Rough Volatility with Backward SPDEs

Path-dependent PDEs for volatility derivatives

Study on the pricing model of path-dependent options

The solutions to a bi-fractional black-scholes-merton differential equation

High-resolution path-integral development of financial options

Option pricing and perfect hedging on correlated stocks

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

Option Valuation using Fourier Space Time Stepping

A Path Integral Approach for Time-Dependent Hamiltonians with Applications to Derivatives Pricing

Option pricing using a skew random walk pricing tree

Multilevel Path Branching for Digital Options

Locally risk minimizing pricing of Asian option in a semi-Markov modulated market

Option pricing models without probability: a rough paths approach

Machine Learning Methods for Pricing Financial Derivatives