A Stock Recommendation Strategy Based on M-LDA Model
Reports from stock analysis are an important source of information in the quantitative investment.How to capture the valuable information from the massive research reports quickly and accurately,and make a good stock recommendation method is one of the important issues in big data quantitative investment.Based on a kind of semi-supervised topic models (M-LDA) and by setting some fundamental emotion labels along with some certain topic labels,we are able to discover the representative words of the research reports for both explicit and latent topics.And then,by calculating the frequency of words in the topics,we designed a stock recommendation strategy.Experiments were carried out the report data from 2011 and 2014 and the results show that this method is effective to find out higher winning-rate stocks.
Min-fan He
School of Mathematics and Big Data,Foshan University,Guangdong 528000,China
国际会议
International Asia-Pacific Web Conference(第18届国际亚太互联网大会)
苏州
英文
122-128
2016-09-23（万方平台首次上网日期，不代表论文的发表时间）