Abstract: At present, the non-financial information with text messages as the principal part is becoming more and more important in annual reports. This paper adopts the textual intonation statistical method and characteristic word statistical method to analyze the text message in the annual reports of China’s A-share listed companies through Python programming language. The findings show that analysts would pay attention to the text information in the annual reports and adjust the performance forecast of listed companies. If the tone of the text message does not match the actual performance and is extremely optimistic, the analyst will downgrade the performance forecast and reduce their focus on the text message.

Key words: text information in annual report; analyst; unusual optimism