||
本文为加拿大多伦多大学(作者:Mohammed Siyam Ibrahim)的学士论文,共122页。
时间序列分析在金融数据中的应用仍然是日内交易员、定量金融专家和精选投资专家感兴趣领域。作者成功地将经典时间序列分析与统计过程控制(SPC)工具相结合,提出了一种能够对多种金融工具的市场表现进行建模和监控的高性能技术分析系统。此外,本文还开发了一种“数据池”技术,推广了自回归系数的标准Yule-Walker方程;应用Box-Jenkins的模型识别、估计和验证方法,基于股票数据的多个非序列历史数据生成ARIMA模型;并比较了使用图形残差分析技术的集合历史模型与传统单一历史模型的准确性。结果表明,虽然集合历史模型通常与单一历史模型同样适用,但前者特别适用于金融时间序列中经常观察到的涉及多个短期历史数据的情况。使用CUSUM控制图监控模型残差提供了一个有价值的概念证明,验证了时间序列分析与SPC工具在建模和监控金融工具行为中的应用。
The application of time series analysis tofinancial data continues to be an area of interest to day-traders, quantitativefinance specialists and select investment professionals. By successfullycombining classical time series analysis with Statistical Process Control (SPC)tools, the authors propose a highly capable technical analysis system whichmodels and monitors market performance of a variety of financial instruments.In addition, the paper develops a ‘data pooling’ technique to generalize thestandard Yule-Walker equations for autoregressive coefficients; applies theBox-Jenkins’ methodology of model identification, estimation and validation togenerate ARIMA models based on multiple non-sequential histories of stock data;and compares the accuracy of pooled-history models to that of conventionalsingle-history models using graphical residual analysis techniques. The resultsindicate that while pooled-history models are generally as adequate as thesingle-history models, the former are particularly well-suited for situationsinvolving multiple short runlength histories which are often observed infinancial time series. Using CUSUM control charts to monitor model residualsprovided a valuable proof-of-concept that validated the use of time series analysisin conjunction with SPC tools in modeling and monitoring the behaviour offinancial instrument.
下载英文原文地址:
http://page2.dfpan.com/fs/1l7cfj1222d1d219169/
更多精彩文章请关注微信号:
Archiver|手机版|科学网 ( 京ICP备07017567号-12 )
GMT+8, 2024-9-27 19:26
Powered by ScienceNet.cn
Copyright © 2007- 中国科学报社