A little while back, researchers at Indiana University – Bloomington caused a mini-uproar by announcing that they had found a way to use Twitter to predict the direction of stocks with startling accuracy. They developed an algorithm that analyzed the sentiment of tweets about various companies, using emotion as the primary indicator of stock performance. It turns out, 86.7 percent of the time they were able to forecast whether a stock was about to rise or fall, a stronger correlation than almost any other tool currently used. This week, a different set of researchers from the University of California – Riverside, alongside a team from Yahoo! was back at it, and, by using a completely different Twitter-based strategy, yielded even more accurate results.
Instead of measuring sentiment, their new strategy involved the pure volume of tweets about a particular company. However, it is more nuanced than basing the strength of stock purely on how much a company is mentioned (if that were the case, time to invest in Tebow Corp). The researchers instead measured the volume of tweets regarding what they termed “connected components,” essentially tweets having to do with important company news (CEO developments, a new product, etc.). The results surpassed other Twitter-based strategies in terms of forecasting stock by between 1.4 and 10 percent.
I find this new study interesting for two reasons. First, it is reasonable to assume that in the near future researchers (perhaps a third group!) will be able to combine the sentiment- and volume-based algorithms to create an extremely powerful stock market forecasting tool that will be the go-to for financial analysts, ushering in a new era in digital economics.
Second, this announcement out of UCAL-Riverside could also go a long way in proving the financial worth of Twitter. Similar to how press releases are keyword-optimized in order to improve SEO, once this research is further developed, companies (probably PR and communications agencies) may be able to figure out what language and which types of news announcements should be aggressively distributed via Twitter in order to improve companies’ forecasts on the market.
Obviously, we are still a long way from being able to use Twitter to analyze the stock market, and future research might show that perhaps the social network is no more precise than fully developed, currently available tools. However, the news out of Riverside looks promising. Twitter has already shown to be valuable from a marketing and PR perspective, is financial analysis the next frontier?
This post was first published by Jason Fidler on March Communications’ blog, PR Nonsense.