Can Profits Still be Made using Neural Networks in Stock Market?

S.-I. Wu and H. Zheng (USA)

Keywords

Neural Networks, Artificial Intelligence, Statistical Modeling, Time Series Analysis

Abstract

In the recent decline of stock market, is it still possible to make some profits? We have developed recurrent neural networks to forecast the daily closing prices of stock indexes (S&P500, Dow Jones, and NASDAQ). Instead of the traditional criteria based on the accuracy of forecasts such as minimum square errors, we chose to use the profit rates as our criteria on evaluating neural networks. Experiments showed that the price is predictable and much better than the random guess. According to a simple short-term investment strategy, a good annual profit rate can be obtained. Different activation functions were tested in order to dynamically choose the best neural network topology. We explored more than 600 network structures for each neural network. The same data sets were analyzed by ARIMA model using commercial statistics package MINITAB. The autoregressive parameter helped us determine the recurrent neural network structure.

Important Links:



Go Back


IASTED
Rotating Call For Paper Image