Artificial intelligence (AI) and machine learning are advancing rapidly and finding uses in many industries from improving production processes to better understanding retail customers. Work by S&P Global Ratings has shown that these methods can now be used to help analyse corporate earnings announcements.
Researchers from S&P Global have used AI and machine learning techniques to scan corporate announcements and conference calls. They are able to use natural language processing to derive scores for sentiment, language complexity and selectivity to give an overall negative or positive view of company announcements.
The sentiment score is based on a formula that looks at the balance between negative and positive words used in the announcement or company briefing. Quite simply, the more negative words used the worse the implications for the company’s outlook.
Language complexity is another tool used by the S&P analysts. They look at factors such as the average number of words used per sentence and the complexity of the word used (number of words with three or more syllables). The logic here is that companies have a duty to disclose negative news to the market but are tempted to use long-winded and complex explanations to try and temper the bad news.
Many market participants are aware that companies often have “favoured” analysts – often those with a positive outlook on the company’s prospects. The AI process can look at the time given to these favoured analysts over others on a call.
As noted above, companies often use longer and more complex answers when giving bad news. The AI driven process is thus able to also pick up on the length of answers compared to averages.
The S&P researchers have found that combining these factors into an overall score provides a useful guide to the likely outlook for a company’s share price.
Business Insider Emails & Alerts
Site highlights each day to your inbox.