Recurrent Support Vector Regression for a Nonlinear ARMA Model with Applications to Forecasting Financial Returns Abstract: Motivated by the recurrent Neural Networks, this paper proposes a recurrent Support Vector Regression (SVR) procedure to forecast nonlinear ARMA model based simulated data...

With the recent availability of high-frequency Financial data the long range dependence of volatility regained researchers' interest and has lead to the consideration of long memory models for realized volatility. The long range diagnosis of volatility, however, is usually stated for long sample...