The Special Aspects of Sentiment Analysis on the French Stock Market

In recent years, due to the growing interest in behavioral finance, the study of the impact of investor sentiment on stock markets has become increasingly in demand. This study is aimed at identifying the features of the influence of sentiment on French stock market returns. The methodology is based on the Shannon transfer entropy, LASSO regression, which were not previously used in the analysis of sentiment regarding the French stock market, as well as econometric regression analysis. For the first time for French stock market, an approach of estimating the effect of investors’ attention on stock market returns through Wikipedia search queries was applied. Several datasets were collected in the period from 2000 to 2023, including data from Twitter (X) and news sites (Bloomberg, Reuters and Investing.com), based on which by using the FinBERT language model to determine the tonality of the text sentiment indices were built. It turned out that investor sentiment, assessed through the tone of messages on Twitter (X), unlike the tone of news headlines in official sources, does not have a significant impact on CAC 40 index. This conclusion differs from other global markets and relates to the specifics of the French market. The paper also identified the distinctive features of the French stock market as a result of influence of investors’ sentiment on stock market returns in the period of economic and geopolitical risks.