Adaptively weighted support vector regression for financial time series prediction

The financial data are usually volatile and contain outliers. One problem of the standard support vector regression (SVR) for financial time series prediction is that it considers data in a fixed fashion only and lack the robustness to outliers. To tackle this issue, we propose the adaptively weighted support vector regression (AWSVR) model. This novel model is demonstrated to choose the weights adaptively with data. Therefore, the AWSVR can tolerate noise adaptively. The experimental results on three indices: the NASDAQ, the Standard & Poor 500 index (S&P), and the FSTE100 index (FSTE) show its advantages over the standard SVR.