Stock price prediction based on a network with gaussian kernel functions

Abstract

This paper presents a new method of stock price prediction based on the phase space analysis for stock price series. For the prediction model, a network with Gaussian kernel functions is selected and this network is optimized by using a noise variance estimate. As a result, the proposed model provides the high accuracy of predicted values. Through the simulation for the prediction of KOSPI 200 stock price values, the effectiveness of the proposed prediction model has been demonstrated. © Springer-Verlag 2013.