Stock price prediction based on robust relevance vector machine with wavelet kernel

In order to solve the QP problem of traditional relevance vector machine effectively, a novel prediction algorithm based on robust relevance vector machine with wavelet kernel (WK-RRVM) is proposed to predict stock price in the paper. In the WK-RRVM model, wavelet kernel is used as the kernel function. The prediction for stock price including opening price, maximum price, minimum price, closing price, trading volume and trading amount of a certain stock based on robust relevance vector machine with wavelet kernel is performed in the study. And the comparison of mean error of stock price including opening price, maximum price, minimum price, closing price, trading volume and trading amount between WK-RRVM and RVM is given. The experimental results show that the prediction results for stock price by WK-RRVM are better than those by RVM. © 2009 by Binary Information Press.