Application And Study On Gene Expression Programming In Stock Prediction

The stock market, which is the main character of market economy, attracts millions of investors from its birth. Within recent ten years, stock has penetrated into our daily life and the population of investing stock has increased to millions. For the enormous profits, study on the internal rule of stock has been a hot topic and there emerge a lot of theories and methods to predict the trend of stock market. However, due to the complexity of the internal structure of the stock price system and the changeability of the external factors, the effect of present prediction methods is not satisfactory.In this thesis, we have done a series of research on Gene Expression Programming, and have applied it to Stock Prediction. Firstly, we described the relevant background and basic principle of stock prediction, discussed some factors which have vital influence to stock price, and summarized some problems in stock prediction fields. Secondly, we described the basic principle of Genetic Algorithms, Genetic Programming and Gene Expression Programming, list Delphi source codes of Gene Expression Programming, and analyzed experimental result which was based on two test problems. Finally, we discussed the selection standard of stock samples, analyzed two common forecast methods, and compared experimental results based on three different stocks.The main contributions in this thesis are listed as follows:1) We analyzed and summarized the basic theories of stock prediction.2) We described the basic principle of Genetic Algorithms, Genetic Programming and Gene Expression Programming.3) We discussed the implementation of Gene Expression Programming, and list its Delphi source codes.4) Experimental results based on two test problems show that Gene Expression Programming can be applied in Function Finding and Data Mining successfully.5) We compared experimental results based on three different stocks, and from experimental data we can find that Gene Expression Programming had a correct rate of 60%-70% when it was applied to short-term stock prediction, and it can give investors some helpful suggestions.