Abstract
It is very useful to utilize the well-performed knowledge by portfolio managers and institutional investors in practical financial dealing. In this paper, we develop a stock investment expert system based on the Granville's law to signal for buy or sell decision, and evaluate quantitatively the predictive performance. Then the fuzzy neural network is used to represent the expert's knowledge on computers and predicts the stock trading price in the market. In empirical experiments, we show that the learning algorithm for the fuzzy neural network can improve the system performance and that the knowledge-based reasoning can catch the nice timing to sell or buy the stock in the fluctuating market.
