Industry Momentum Strategy Based on Text Mining in the Japanese Stock Market

Abstract

Industry momentum refers to the phenomenon that buying(selling) industry portfolio from the past winner(loser) generates a positive return. In general, there is some lead-lag effect in the spread of price information due to the limited attention of investors. The industry momentum exploits the lead-lag effect due to information spillover delays within industries. In this study, We test whether industry momentum is stronger when we use information that is not attracting the attention of investors. We first perform industry classification with text mining of Annual Securities Reports describing the business activities of each company listed on the Japanese stock market. Then, we can identify groups of industry peer companies with low investor attention that is likely to cause the lead-lag effect. Such a text-based industry classification based on Annual Securities Reports is less visible to the investors but has an economic link among companies. We confirm that industry momentum base on text-based industry classification is more significant than traditional industry classification in Japanese stock market.