2023 Australasian Actuarial Education and Research Symposium

Erwinna Chendra

Parahyangan Catholic University

Valuing an arithmetic Asian option with Artificial Neural Network method

This is joint work with Christian Jauhari, Andreas Parama Wijaya

Pricing an Asian option with an arithmetic average is interesting because there is no analytical solution. Usually, pricing such options uses numerical methods, one of which is the Monte Carlo method. This method's weakness is the computation time length because it must generate many trajectories of asset price movements. This paper uses a data-driven approach through the Artificial Neural Network (ANN) method that can speed up computation time. The ANN method requires input and target data sets from Lévy's approximation solution, Monte Carlo with antithetic variables, and Quasi-Monte Carlo. The ANN method learns the data from these three data sets to form each model that can accept the input of option parameters to determine the price. The most crucial thing in ANN models is the model architecture, such as assessing the number of neurons, layers, and the use of activation functions. This paper presents several simulations to determine the best model architecture based on the smallest Mean Squared Error (MSE). In addition, the evaluation of the three ANN models that have been built shows that each model is optimal based on the smallest MSE and a coefficient of determination close to one. This paper also shows that option pricing using the ANN method is more efficient than the Monte Carlo method.