Wavelet-based relevance vector machines for stock index forecasting

Relevance vector machine (RVM) is a Beyesian version of the support vector machine, which with a sparse model representation, has appeared as a powerful tool for time series forecasting. RVM has demonstrated better performance over other methods such as neural networks or ARIMA-based models. This paper proposes a wavelet-based RVM model to forecast stock indices. The time series of explanatory variables are decomposed by the wavelet basis, and the extracted time scale features served as inputs of a RVM to perform the nonparametric regression and forecasting. Compared with the traditional GARCH model forecasts, the new method shows superior performance, and reduces the root-mean-squared forecasting errors by nearly one order.