Twitter useful for stock picks, could boost market efficiency, shows study
Despite concerns to the contrary, it turns out that Twitter is not uninformative when it comes to its reliability as an investor information source.
The study was motivated by exponential growth in "peer-to-peer" sharing and information tools, exemplified by social media such as Twitter, said Partha Mohanram, John H. Watson Chair in Value Investing, and professor of accounting at the University of Toronto's Rotman School of Management. He conducted the study with researchers from two other universities.
"If you want to find out if a stock or company is likely doing well or doing badly, you don't just follow what the person in The Wall Street Journal or Barron's magazine is saying," said Prof. Mohanram, who researches valuation, "you also focus on what other people who are also investing in stock markets are saying."
The study applied textual analysis techniques to 870,000 tweets about 3,600 companies between January 2009 and the end of September 2012. On the whole, tweets in the days leading up to a firm's quarterly earnings reports successfully predicted whether the firm would meet its targets or not. Twitter activity also did well in forecasting how the stock price would subsequently respond.
The effect was especially strong in situations where there was generally scant information about a company, such as a small firm with limited marketing and communications capacity, and for tweets that conveyed information specifically about a firm's fundamentals and stock activity. Cases like those underscore how social media can promote market efficiency by sharing additional valuable information.
"The tweets are from both people expressing their own original opinions about firm fundamentals and firms passing on existing information," said Prof. Mohanram. "This highlights the dual role of Twitter—as a source of new information and as a new mechanism to disseminate existing information.
The results reflect the concept of the "Wisdom of Crowds," where it's been found that a large group of people en masse can better predict a result than a single expert.
Although Twitter has been misused to communicate damaging, false reports about a firm, the researchers say their study shows why regulation to reduce such misleading information is not warranted. The diversity and independence of opinion provided through social media should trump concerns about limited cases of manipulation, they say.
More information: Eli Bartov et al, Can Twitter Help Predict Firm-Level Earnings and Stock Returns?, The Accounting Review (2017). DOI: 10.2308/accr-51865
Provided by University of Toronto