An Explicit Feature Selection Strategy
About this An Explicit Feature Selection Strategy
The focus of this study is the selection of an appropriate set of features for a feed forward neural network model used to predict both future market direction and future returns for the S&P 500 Index. The experimental results provide evidence that the proposed feature selection process may result in a more successful prediction model. However, the study also indicates that the problem domain may need to be limited to predicting monthly instead of daily movements. In addition, the proposed process could be more useful for predicting the future market direction rather than actual returns. 1. Introduction While the application of neural networks to financial forecasting is beginning to receive academic attention [Freedman 1995], the issue of feature selection for financial forecasting problems has been largely ignored.
Author: Tim Chenoweth
Tim Chenoweth is an Associate Professor of Information Technology at Boise State University. Before receiving his Ph.D. from Washington State University (1996), he was an active duty officer in the U.S. Coast Guard (1981-1989). His assignments included deck watch officer aboard U.S. Coast Guard Cutters Cambell and Yocona, IT department Coast Guard 11th district Headquarters, Long Beach Ca., and head of IT, Civil Engineering Department, Coast Guard Headquarters, Washington D.C. Prior to arriving at Boise State, he was an Assistant Professor in the Computer Information Systems department at Arizona State University (1996 to 2003).