Font Size: a A A

Research On Investment Strategy Of Public Opinion Tracking Based On Machine Learning

Posted on:2021-04-04Degree:MasterType:Thesis
Country:ChinaCandidate:W YuFull Text:PDF
GTID:2428330647950693Subject:Integrated circuit engineering
Abstract/Summary:PDF Full Text Request
Because of its low barriers to entry and high profit cap,stock market investment has been a hot research topic for a long time.China's stock market is a market dominated by individual investors.Through the real-time tracking of public opinion related to the stock market,the research of the emotional tendency contained in it can insight into the psychological state of shareholders in a sense.So in a theory,a set of investment strategies based on public opinion can be constructed for practical transactions.This paper proposes a new investment strategy based on public opinion tracking using machine-learning skills.First,by recording the real-time search results of specific keywords on the third-party platforms such as Baidu Index,We Chat Index,Micro-Blog Index,etc.,the temperature of the market can be perceived.Furthermore,by collecting the posts in the stock forum,the daily index of stock review sentiment can be defined and generated and the human nature can be grasped.Through the combination of crawler and optical character recognition,the method proposed in this paper greatly improves the stability of data collection system.In order to overcome the accuracy problem of existing algorithms in text emotion recognition,careful selection of training samples and design of network structure are performed to gives full play to the advantages of different machine learning models,and ensemble learning ideas are used to carry out emotion prediction.Compared with the existing algorithms,the model proposed in this study can achieve higher emotion classification accuracy in the text.Finally,combined with the collected public opinion data,a variety of trading strategies based on public opinion are generated.After the backtracking test on the historical market data,the transaction model based on the comprehensive public opinion index reversal strategy which performs the best is selected.The experimental results show that compared with the comparative benchmark,it has fairly good ability to obtain excess return while resisting risk,which verifies the effect of the public opinion strategy for actual investment.
Keywords/Search Tags:Stock sentiment, Emotion recognition, Machine learning, Tradinstrategy
PDF Full Text Request
Related items