40 Neural Based Technical Analysis in Stock Market Forecasting
Download citation file:
In this study, a Jordan∕Elman neural network model is proposed for stock market forecasting using technical trading indicators in order to predict the next day closing price of a particular stock value. The technical indicators chosen to represent the stock characteristics are sampled from independent indicator types, such as trend, mode, momentum and volatility. The training data is obtained from NYSE historical data of TKC stock and QQQQ ETF between 1999 and 2005. With the trained network, 2006 data has been tested and promising results are achieved. For performance evaluation, different buy-sell rules are created using the most common technical indicators and the rate of returns are compared with the neural network forecaster's performance. The results indicate that the neural network prediction model outperformed all of the other models. This system can be used for the management of a combination of stocks and ETFs providing short term buy-sell signals.