Associating stock prices with web financial information time series   based on support vector regression

doi:10.1016/j.neucom.2013.01.011

CORC > 北京大学 > 信息科学技术学院

	Associating stock prices with web financial information time series based on support vector regression
	Liang, Xun ; Chen, Rong-Chang ; He, Yangbo ; Chen, Ying
刊名	neurocomputing
	2013
关键词	Web financial information time series Stock price time series Support vector regression TRADING VOLUME MODEL VOLATILITY MARKETS TEXT
DOI	10.1016/j.neucom.2013.01.011
英文摘要	Over the past years, the effect of massive information from the web on the financial market has increased. How to process and utilize such information attracts both researchers and portfolio managers. In this paper, financial information obtained daily from the web is treated as a time series and then associated with stock price volatilities. First, six research topics on financial time series are outlined, namely, analyses of stock price time series P, trading volume time series V, web information time series W, and relationship between P and V, P and W, as well as V and W. Second, a model connecting P and W based on the support vector regression (SVR) is examined as an example of the six research topics. Third, given that a typically successful way of computer-based natural language processing is through the conduct of keyword analysis, the novel finance-computer time series W is explicitly defined in terms of financial keywords and is used in the present paper as the topic of investigation. The relationship between P and W is modeled using SVR. Because during the pre-web era people cannot manually and efficiently process image information from the newspapers and sounds from the television and radio over a longer period of time (e.g., a year), they were unable to obtain the time series W. Therefore, it is the web that makes the research on the relationship between P and Win the meaning of quantity of W possible. Finally, experiments on the Shanghai and Shenzhen security markets revealed that the introduction of W helps improve model accuracies. As the web further develops, more and more ordinary people share their views on the web. The "long-tail" of massive financial information formed by these "grass roots" has a noticeable effect on the financial markets. In financial markets, those who quickly capture and interpret financial information have the potential to generate profits. With the use of the newly found model connecting P with W and fast decision making, financial market practitioners can be rewarded. (C) 2013 Elsevier B.V. All rights reserved.; http://gateway.webofknowledge.com/gateway/Gateway.cgi?GWVersion=2&SrcApp=PARTNER_APP&SrcAuth=LinksAMR&KeyUT=WOS:000320476500014&DestLinkType=FullRecord&DestApp=ALL_WOS&UsrCustomerID=8e1609b174ce4e31116a60747a720701 ; Computer Science, Artificial Intelligence; SCI(E); EI; SSCI; 7; ARTICLE; 142-149; 115
语种	英语
内容类型	期刊论文
源URL	[http://ir.pku.edu.cn/handle/20.500.11897/161771]
专题	信息科学技术学院
推荐引用方式 GB/T 7714	Liang, Xun,Chen, Rong-Chang,He, Yangbo,et al. Associating stock prices with web financial information time series based on support vector regression[J]. neurocomputing,2013.
APA	Liang, Xun,Chen, Rong-Chang,He, Yangbo,&Chen, Ying.(2013).Associating stock prices with web financial information time series based on support vector regression.neurocomputing.
MLA	Liang, Xun,et al."Associating stock prices with web financial information time series based on support vector regression".neurocomputing (2013).