| Dow theory considers that stock market changes regularly as well as the development of economy. This changing rule makes the stock market move to some direction, which can be seen from the movement of the price of stock. This paper which is based on Dow theory, forecasts the movement trend of the stocks and gives available trade strategy by using mathematical analysis, fuzzy decision and computer software technology.At present, stock analysis system mainly forecasts using nerve net, which trains andante and there is least point in object function. This paper brings forward a new method to establish a fuzzy expert system of stock analysis system, which contains explaining interface, fuzzy inference, learning module, fuzzy knowledge library and memorizer. The learning module concludes fuzzy rules, summarizes 13 kinds of characteristic point from the swatch room, calculate the rising probabilities of samples behind each kind of characteristic point, then average every kind of rising probabilities of samples to 13 rule tables. Inference turns the fuzzy rule to accurate probability using Mamdani direct inference. Then the explaining interface turns the accurate probability to trade strategy using Max-Min inference. Thus a trade strategy is given by disposing history data and forecasting the movement trend of stock.This paper has 8 chapters. In the first chapter, the current stock analysis system is introduced. In the second chapter, Dow theory is introduced. In the third chapter, the fuzzy decision system is introduced. In the fourth chapter, the candlestick charts is introduced and the 13 kinds of characteristic point are explained. In the fifth chapter, every structure of the fuzzy expert system of stock analysis system is illuminated. In the sixth chapter, arithmetic of each structure is illuminated. In the seventh chapter, the software is showed with flow chart. In the eighth chapter, the result of experiment is given to prove the feasibility of this system. |
