Sentiment in Financial News and the Application of Portfolio Management

Abstract

We construct a sentiment index with the incorporation of representative proxies in the Taiwan Stock Market and investigate the application of a cross-section of stock returns. The major contribution of the study is that the huge Chinese financial news related to each listed stock is collected and the market aggregate news confidence index (ANCI) is constructed using linguistic analysis. The empirical results show that the monthly sentiment in financial news could be a leading index of the market returns. However, the investors’ fear gauge of the volatility index decreases when the lagged term of ANCI increases. The analysis supports the view that ANCI could be included in alternative sentiment proxies by examining the principal components analysis referred to in Baker and Wurgler (2006). The application of portfolio analysis confirms that when the sentiment in the previous month is low, subsequent value-weighted portfolio returns are relatively high with lower turnover, larger size, higher market value to book value and older age. When sentiment is high, the subsequent returns are relatively low regardless of whether they are equal-weighted or value-weighted. This study contributes to previous studies which support the view that sentiment in financial news should be incorporated into one of the sentiment proxies for decision making.

Keywords: Sentiment, News, Stock Returns, Linguistic Analysis, Portfolios

JEL Classification: D82, G12, G14