時間序列波動特徵研究

CNU IR > College of Smart Living > Dept. of Information Management > CNU Project > Item 310902800/9923

請使用永久網址來引用或連結此文件: https://ir.cnu.edu.tw/handle/310902800/9923

題名:	時間序列波動特徵研究
作者:	蘇致遠
貢獻者:	資訊管理系
關鍵詞:	Multifractal Stock Market Econophysics
日期:	2008
上傳時間:	2009-02-20 11:44:04 (UTC+8)
出版者:	台南縣：嘉南藥理科技大學資訊管理系
摘要:	This paper analyzes the minute-by-minute variations of the Taiwan Stock Exchange Capitalization Weighted Stock Index (TAIEX) over eight years using box-counting multifractal spectra . The results reveal that the daily return R is directly correlated with the absolute value of for that day, while a positive or negative sign of is related to an increasing or decreasing return, respectively. The gain probability (G%) and index increase probability (N%) attain 65~74% when has a positive value and 8~32% when has a negative value, but both converge toward 50% as the number of days considered when computing the value of increases. With regard to prediction of the future index movement, the results show that the sign sequences of provides a more reliable predictive performance than that of the index variation parameter . The correlation between the risk measurement parameter and the increasing or decreasing tendency of the TAIEX price index is also examined in this paper, and results are opposite to those presented for the SSEC index in China. It is thus suggested that the phenomenon is market dependent.
關聯:	計畫編號：CN9733
顯示於類別:	[Dept. of Information Management] CNU Project

文件中的檔案:

檔案	描述	大小	格式	瀏覽次數
97CN9733.pdf		379Kb	Adobe PDF	765	檢視/開啟

在CNU IR中所有的資料項目都受到原著作權保護.

TAIR相關文章