[2A6-GS-6] Language media processing: applications

[2A6-GS-6-01] Extraction of theme stocks using individual stock information and cross sectional stock information

〇Takehiro Takehiro Takayanagi¹, Hiroki Sakaji¹, Kiyoshi Izumi¹ (1. University of Tokyo)

Keywords:Embedding

In this research, we propose a method to extract stocks related to a certain theme by using stock embeddings. Various investment trusts treat thematic mutual funds. Thematic mutual funds are created and sold according to the trends of the market and have been increasing in popularity, especially among Japanese retail investors. However, the stocks related to the theme are manually selected and it is costly to build. Therefore, in this research, We propose a method to help extract thematic mutual funds by embedding stocks’ individual information and cross-sectional information. As a result of evaluation experiments, our proposed method is better than the comparative method. It was shown that stocks can be extracted with high performance.