Stock Returns and Investor Sentiment: An Empirical Study of the French Stock Market

Abstract

Using computational linguistic techniques, we build an investor sentiment indicator extracted from the content of the electronic French press specialized in Financial and economic news for companies of the CAC40 index. We test the relationship between this indicator and abnormal returns estimated at the opening of the stock market for each security. Using Granger causality tests, we find that the investor sentiment indicator determines abnormal returns and not the opposite. Multiple linear regression results show also that it can predict abnormal returns at the opening of the next day. Finally, we test an investment strategy based on sentiment values and achieve a better performance than the CAC40 Index including transaction costs.