科研之友
微信
新浪微博
Facebook
分享链接
摘要
Traditional stock analysis is data mining, based on the stock price history, such that the future stock trend can be predicted. In those approaches, the stock price history was usually considered as a time series. However, the prediction based on only a few past data in the stock time series often cannot provide satisfactory results. Technical analysis is a commonly used tool in predicting stock trends. The technical indicators are coupled and may conflict with each other. In this research, we employed model-free estimators to model intricate relationships between technical indicators and stock trends. Five technical indicators are used as input variables and the price difference is used to define the desired outputs, which are used to form stock trade strategies. In this paper, a way of transforming a stock time series into meaningful input variables based on the technical indicators is proposed. A fuzzy mechanism is also included to define the desired operational signals from the stock trend. The results have demonstrated efficiency when compared with the direct use of the technical indicators. Furthermore, those indicators cannot completely define the outputs because there are also other factors that may influence the stock price. Thus, the relationship between the input and the output data is not deterministic. In order to consider such a non-deterministic property, an approach that makes use of the Self-Organizing Map clustering technique is proposed, to measure the degree to which the data is non-deterministic. This degree is then modeled into a network and is used in the decision-making process. The results of this algorithm are indeed superior to those of the original approach.
- 出版日期2003-1
