Combing extended Kalman filters and support vector machines for online option price forecasting

Abstract

This study combines extended Kalman filters (EKFs) and support vector machines (SVMs) to implement a fast online predictor for option prices. The EKF is used to infer latent variables and makes a prediction based on the Black-Scholes formula, while the SVM is employed to capture the nonlinear residuals between the actual option prices and the EKF predictions. Taking option data traded in Taiwan Futures Exchange, this study examined the forecasting accuracy of the proposed model, and found that the hybrid model is superior to traditional feedforward neural network models, which can significantly reduce the root-mean-squared forecasting errors.