遺傳規畫法於股價預測模組之應用

The Application of Genetic Programming on the Stock Movement Forecasting System.

蔡奕祺(Yi-Chi Tsai)

指導教授 ： 周宗南 ； 洪振義

朝陽科技大學/管理學院/財務金融系/碩士(2010年)

摘要

This paper intends to demonstrate the application of an artificial intelligence system named GPLAB on the prediction of stock price movement in TWSE.
GPLAB was built on biological evolutionary concept to realize fittest surviving rules in the natural selection process. This concept has been applied on the field of finance to build up forecasting models predicting future price movement within one day, one month and one season.
The empirical results of this inter-discipline study has revealed this bio-financial hybrid system successfully predicted the stock price movement in a one-month forecasting range by 23% and 22% lower than the appointed benchmark during a random chosen period and a bear market period respectively. This empirical evidence suggests the market efficiency in TWSE is a semi-strong form market that stock price movement could be predicted with the analysis of historical data.
This paper also further indicates the credibility of different technical and fundamental factors regarding to the prediction of future price movement in four different market situations including non-specific, static, bull and bear market period.
At the end of this paper also revealed the strength and weakness of GPLAB as a financial forecasting tool. A short discussion concerning the system improvements regarding to the application of GPLAB is also included.

關鍵字

股價 ； 預測系統 ； 遺傳規劃法

並列摘要

This paper intends to demonstrate the application of an artificial intelligence system named GPLAB on the prediction of stock price movement in TWSE.
GPLAB was built on biological evolutionary concept to realize fittest surviving rules in the natural selection process. This concept has been applied on the field of finance to build up forecasting models predicting future price movement within one day, one month and one season.
The empirical results of this inter-discipline study has revealed this bio-financial hybrid system successfully predicted the stock price movement in a one-month forecasting range by 23% and 22% lower than the appointed benchmark during a random chosen period and a bear market period respectively. This empirical evidence suggests the market efficiency in TWSE is a semi-strong form market that stock price movement could be predicted with the analysis of historical data.
This paper also further indicates the credibility of different technical and fundamental factors regarding to the prediction of future price movement in four different market situations including non-specific, static, bull and bear market period.
At the end of this paper also revealed the strength and weakness of GPLAB as a financial forecasting tool. A short discussion concerning the system improvements regarding to the application of GPLAB is also included.

並列關鍵字

Stock Movement Forecasting ； GP ； Genetic Programming

參考文獻

29. Modigliani, Franco, Merton H. Miller, (1958). The cost of capital, corporation finance and the theory of investment, The American Economic Review 48, 261-297.

19. Gallant,Ronald A., Peter E. Rossi, George Tauchen, (1992).Stock prices and volume, The Review of Financial Studies5, 199-242.

1. Banz, Rolf W. (1981). The relationship between return and market value of common stocks, Journal of Financial Economics 9, 3-18.

2. Basu, Sanjoy (1983). The relationship between earnings’ yield, market value and return for NYSE common stocks, Journal of Financial Economics12 ,129-156.

4. Birchenhall,C. R. (1995). Genetic algorithms, classifier systems and genetic programming and their use in the models of adaptive behaviour and learning, The Economic Journal105, 788-795.