Intraday volatility forecasting for option pricing using a neural network approach

Good implied volatility estimates are required to correctly evaluate financial options, forcing option market participants to look for a method to measure it. Due to the intrinsically nonlinear features of implied volatility measures, nonlinear approaches are necessary to model it. We propose an integrated modelling strategy that makes use of a nonlinear general function approximator, the artificial neural model (ANN) and classical linear techniques. This modeling strategy departs from the least available information given by the univariate analysis of the output series. From this bottom line we enrich our modelling with multivariate information: first, making use of standard econometric linear methods and then embedding the information obtained in this step of the process in a more complex and non-linear model, the ANN.