Bullish-bearish-based neural network stock trading decision supportano its application in Hong Kong stock market

Stock trading is an important financial activity. Accurate prediction of stock performance may yield a large amount of profits. However, the stock price is influenced by many different factors and there are many investors involved, the Efficient Market Hypothesis shows that it is impossible to predict the stock price fast enough to make profit. Instead, one makes profit by correctly predicting the trend of stock, instead of the accurate stock price at a particular time slot. Therefore, this work proposes a stock trading decision method based on candlestick patterns and a Multiple Classifier System (MCS) consisting of several Radial Basis Function Neural Networks (RBFNN) trained via a minimization of the Localized Generalization Error Model (L-GEM). The major contribution of this work is a new buy-and-sell strategy which is more reasonable than the 3-day trading strategy in our previous works. Experimental results show that the proposed strategy yields better results and yields higher profits.