Pricing European Options with Deep Learning Models

Applications of artificial neural networks for valuing financial options have a long history in academic research. When appropriately trained, these artificial neural networks have proved to obtain better results than traditional parametric models like Black-Scholes. This short paper intends to review the input variables and specific architecture of a multi-layer perceptron, long-short term memory and a convolutional neural network model for the purpose of pricing options. Additionally, it describes the data gathering process for the dataset that will be used to train these models.