THB, Banditos, Wayward and more confirmed for Cosmic Cocktail!

Hopkins research finds Twitter mood predicts success of IPOs

Want to know if a company about to go public will succeed? Check Twitter, Hopkins study says.

Wondering whether to take a chance and invest in a company that's about to go public?

You might want to tune in to Twitter, researchers at the Johns Hopkins University say.

The chatter on Twitter and other social media channels is already being mined to check the pulse of movie premieres, new product launches and even the broader stock market. At Hopkins, a finance professor and graduate student have zeroed in on Twitter's impact on a more focused equity arena — initial public offerings, or the first time a company sells stock and it begins trading on a public exchange.

"Twitter matters for financial data," and can be used to predict IPO performance, researcher Jim Kyung-Soo Liew said he concluded after completing the study, released last week. "The sentiment obtained from tweets … matters for IPOs and speculative securities in the early stage."

Liew, an assistant professor at Hopkins' Carey Business School, said the project seemed relevant at a time when the IPO market is hot and Twitter usage continues to grow.

But "this relationship between Twitter sentiment and security prices, it's a complicated relationship," he said. "We're seeing evidence whereby investors have potentially overreacted" in their tweets in the days leading up to an IPO.

The study examined the mood of tweets associated with all 325 IPOs that took place in 2013 and 2014. Some of last year's biggest IPOs included mobile coupon company Coupons.com, online food ordering service GrubHub, Lending Club and the biggest IPO ever — Chinese online marketplace Alibaba, which raised $22 billion.

All of those stock sales were talked about on Twitter, which has nearly 290 million active monthly users who send out about a half-billion tweets every day. Investors are increasingly turning to social media sites to analyze and discuss the market, even in Twitter's limited 140-character posts.

The Hopkins study examined whether any relationship existed between the mood of related tweets and how the stock performed on its first day of trading.

To quantify "tweet sentiment," the study relied on data from social media analytics firm iSentium, which converts stock-related tweets into scores, from minus 100 for the most negative to plus 100 for the most positive.

Despite the fact that "tweets are generated by anybody," Liew said, and "the users are not necessarily all financial professionals," the research found a positive relationship. That means that positive sentiment drove up the price and negative sentiment pushed it down.

But researchers also found that Twitter sentiment during the three-day run-up to the stock sale had an inverse relationship to how the IPO unfolded. Opening day prices generally fell after days of positive tweets, while prices typically increased after days of gloomy tweets.

Positive sentiments showed that "they're overly optimistic about the performance of IPOs," Liew said. "If you see an IPO stock and a lot of positive sentiment about this particular stock leading up to the first open, you have to be a little careful about trying to buy that. The market might be overreacting and the price may be too high. The flip side is if there's a lot of negative tweets, you may benefit by purchasing that stock."

Those findings came as no surprise to Richard L. Peterson, a psychiatrist and CEO of MarketPsych Data LLC, a behavioral finance consulting firm.

"People get excited about something, and they tend to overreact," Peterson said. "The excitement builds up to the day of the event, and you have a potential for a big reversal. You see that around Apple products.

"What you generally find is the more people talk about the price going higher and higher, the more the frenzy builds, the more the price will likely fall."

In the Hopkins study, that trend held true with companies such as GrubHub Inc. and Paylocity Holding Corp., a payroll software company, in which positive pretrading sentiment preceded negative opening day returns, said Garrett Zhengyuan Wang, the graduate student who researched the project.

GrubHub, an online restaurant delivery service, sold at $26 a share and opened trading on April 4, 2014, at $40 a share before falling to close at $34 each. Paylocity shares sold for $17 each and opened March 19, 2014, at $31 a share before falling to close at $24.04 each.

On the other hand, strong negative sentiment on Twitter over three days led to positive first-day returns for Kindred Biosciences Inc., a biopharmaceutical company that develops treatments for pets, and specialty grocer Sprouts Farmers Market Inc., Wang said.

Kindred sold its stock at $7 a share and it opened trading Dec. 12, 2013, at $8.75 a share before climbing to close at $11.95. Shares in Sprouts sold at $18 a share, but opened at $35 a share on Aug. 1, 2013, rising to close at $40.11.

Peterson said broad anger can be a good predictor of a stock's positive performance, while joyful comments can signal a price drop. His firm measures emotions such as joy, fear, doubt and anger as well as overall sentiment related to stocks from a range of social media — blogs, tweets and message boards, for example — and provides aggregate data to clients such as banks, hedge funds, economic research departments and governments.

"You actually see that the brain activates with excitement before the event," he said. "It gets us excited to where we're taking risk and that turns off the risk detection areas of the brain. We lose the ability to detect that there might be risk. That leads us to take more financial risk. Then we wake up and think, 'What have I done?'"

Most people who use Twitter to help them trade do so on their own, Peterson said.

"It's only recently that data like ours is available," he said. "Now, instead of reading tweets, they can look at numbers."

The Hopkins study adds to a body of research about how social media can be used as an indicator. Several studies have already shown that Twitter mood can predict the stock market.

In 2010, researchers at HP Labs in Palo Alto, Calif., extracted nearly 3 million tweets referring to 24 movies released over three months. They showed that tweet sentiment can forecast box-office revenues.

Liew said he believes researchers are only beginning to uncover the ways in which social media data matters in finance.

"It is gaining enough traction and people are using it, and the crowd is telling us something," he said.

Copyright © 2019, The Baltimore Sun, a Baltimore Sun Media Group publication | Place an Ad
46°