On Tuesday, April 11th, 2006, Doctor Chen Shiyi from China Centre for Econo-mic Studies Fudan University presented his paper " Recurrent Support Vector Reg-ression for a Nonlinear ARMA Model with Applications to Forecasting Financial Returns". Doctor Lu Ming, Doctor Chen Zhao, Doctor Jiang Jianqiang, Doctor Fan Xiangyan attended the lecture.
At the beginning of the report, Doctor Chen gave a brief in-troduction of the theory of SVR. As opposed to traditional methods and ANN, this thesis applies the predictably powerful SVR to im-proving the forcasting ability of time series models. Then Doctor Chen described the main advantage and the goal of SVR. Also he reviewed the literature on this field. As the report come to the main part, Doctor Chen illustrate the idea and procedure of app-lying RSV to forecasting financial return. By using the real data to make a simulation, we can saw the recurrent SVR has sup-erior forecasting performance to the competing method. At last Doctor Chen presented the following conclusion:(1) The evidence from both Monte Carlo Simulation and the real data analysis con-sistently reveals that the proposed recurrent SVR significantly improves the generalization ability and can be used as a standard procedure in other applications.(2)Sensitivity of recurrent SVR results to kernel type and other free parameters were also examined.
After the report of Doctor Chen, All the teachers and students made a discussion about the interesting model.
