Paper Title
Using A New Artificial Neural Network Algorithm for Predicting Stock Market

Abstract
In the past decade, several researchers worldwide have applied artificial neural networks (ANNs) for stock market prediction, bankruptcy, and prediction. Some of the researchers such as; Kimoto et al. (used modular neural networks for the prediction of the Tokyo stock exchange prices index (TOPIX)), Kamijo and Tanigawa (used recurrent neural networks), Ahmadi use backpropagation neural networks, Yoon and Swales used qualitative and quantitative data, Trippi and DeSieno,and Choi et al. predicted the daily direction of the change in the S&P 500 index futures using ANNs. These studies used different ANN methods for stock market prediction. Most recently, some researchers such as; Tsaih et al., Kohara et al have started using hybridize several artificial intelligence (AI) techniques to improve the prediction performance. The present study uses a new algorithmic method based on value trading to identify when to place buy, sell, or stop orders. After classifying the orders into those three categories, an Artificial Neural Network (ANN) with three layers, input, hidden, and the output is used to learn from previous trades and predict for one day ahead. In order to conduct this research, the value-weighted and equal-weighted indices are downloaded from the CRSP (Center for Research in Security Prices) database. The daily price between the dates 5/26/1927 and 5/25/2007 including 9-day EMA, 18-day-EMA, MACD (Moving Average Convergence Divergence), MACD-EMA (Moving Average Convergence Divergence- Exponential Moving Average), and MACD-sign (Moving Average Convergence Divergence-sign), minimum price and maximum price are chosen as input for training the classifiers. The output of the classifier is one of the labels BUY, SELL, or STOP. The data between the dates 5/26/2007 and 5/26/2017 are used to test the classifiers. This period includes a market crash that happened in May 2010 and August 2015. The goal is to evaluate the performance of the classifier in that period to see whether it can predict the market crash. The learning performance measures including accuracy, precision, recall, and F-score are measured. Keywords- Artificial Neural Network, Stock Market Prediction, Feed-forward Learning.