2019/06/21 by Joshua Zoen Git Hiew, Hiew, Joshua Zoen Git, Xin Huang +9
Computer Science · Decision Sciences · Economics, Econometrics and Finance · Materials Science · #FOS: Economics and business #Financial Markets and Investment Strategies #General Finance (q-fin.GN) #Machine Learning in Materials Science #Sentiment Analysis and Opinion Mining #Statistical Finance (q-fin.ST) #Stock Market Forecasting Methods
paper · pdf · doi:10.48550/arxiv.1906.09024
openalex publication_date 2019/06/21 · openalex created_date 2021/02/01 · openalex updated_date 2026/07/28
Traditional sentiment construction in finance relies heavily on the\ndictionary-based approach, with a few exceptions using simple machine learning\ntechniques such as Naive Bayes classifier. While the current literature has not\nyet invoked the rapid advancement in the natural language processing, we\nconstruct in this research a textual-based sentiment index using a well-known\npre-trained model BERT developed by Google, especially for three actively\ntrading individual stocks in Hong Kong market with at the same time the hot\ndiscussion on Weibo.com. On the one hand, we demonstrate a significant\nenhancement of applying BERT in financial sentiment analysis when compared with\nthe existing models. On the other hand, by combining with the other two\ncommonly-used methods when it comes to building the sentiment index in the\nfinancial literature, i.e., the option-implied and the market-implied\napproaches, we propose a more general and comprehensive framework for the\nfinancial sentiment analysis, and further provide convincing outcomes for the\npredictability of individual stock return by combining LSTM (with a feature of\na nonlinear mapping). It is significantly distinct with the dominating\neconometric methods in sentiment influence analysis which are all of a nature\nof linear regression.\n