摘要
描述了从股票价格时间序列中检测出一些对技术分析有用的特定的几何图形(如双顶,双底,头肩顶)的算法及其实现。首先得到用户指定的股票和时间段内的价格时间序列(这里是每天的收盘价格),然后从中得到一部分直观上对技术分析重要的点,最后根据这些点来判断整个时间序列中是否有符合用户指定的几何图形。该方法用Java实现,并使用JFreeChart工具绘制结果。实践证明该方法能有效地识别几何图形。
This paper presents a method and its implementation which detects a certain set of geometrical patterns （ such as double top, double bottom,head-shoulder-top） that are important to stock technical analyzing from stock price time series. The method first gets the price time series （ here we use the elosiing price everyday for a stock） and then extracts some perceptually important points which are used to determine whether a certain geometrical patterns exist in the series. The method is implemented using Java programming language and the result is presented with the help of the tool JFteeChart, The results proved that this method is effective in the recognition of geometrical pattern from price time series.
出处
《计算机应用与软件》
CSCD
北大核心
2007年第1期175-176,187,共3页
Computer Applications and Software
关键词
股票价格
时间序列
检测几何图形
Stock price Time series Detecting geometrical pattern
