%0 Journal Article %A Mike Chen %A Jaime Lee %A George Mussalli %T Teaching Machines to Understand Chinese Investment Slang %D 2020 %R 10.3905/jfds.2019.1.020 %J The Journal of Financial Data Science %P 116-125 %V 2 %N 1 %X The Chinese A-shares equity market is characterized by heavy retail investor participation. Furthermore, much of the Chinese retail trading is sentiment driven. In this article, the authors discuss their approach of combining a recently developed natural language processing (NLP) technique with online Chinese-language investment blogs favored by Chinese retail investors to understand their views toward various A-share stocks. In addition to complexities associated with NLP application to the Chinese language, one particular challenge posed by the retail investor corpus the authors use is the frequent occurrence of slang. They demonstrate that their approach is able to overcome the challenges encountered. The authors also show how modifying the standard neural network underpinning their NLP approach makes it more suitable for application to the financial domain.TOPICS: Equity portfolio management, statistical methods, simulations, big data/machine learningKey Findings• To understand Chinese A-share sentiments, it is essential to understand the sentiment of retail investors.• The Chinese language poses several distinct challenges that make application of traditional NLP techniques not particularly fruitful.• Using the latest neural network–based NLP approach, we can overcome these challenges and gain insights into the A-shares market. %U https://jfds.pm-research.com/content/iijjfds/2/1/116.full.pdf