Two researchers at Johns Hopkins University, who have studied the effects of how Twitter sentiment impacts the performance of initial public offerings have found what appears to be a significant relationship between the mood of the tweets and the performance.

Twitter sentiment during the three-day run-up is usually opposite to how the IPO will unfold.
In the past, researchers have noted that Twitter sentiment— collective attitude about a topic, as expressed on the popular social media platform—is a reliable forecaster of events. Studies published in recent years, for example, have shown tweets to be a predictor of how stocks will fare from day to day on Wall Street or how movies will play on Main Street.

The new Johns Hopkins University study by Assistant Professor Jim Kyung-Soo Liew and finance graduate student Garrett Zhengyuan Wang, both of the Johns Hopkins Carey Business School, is believed to be the first to look closely at the connection between Twitter sentiment and IPOs.

The two researchers examined data from January 2013 through December 2014 about Twitter sentiment during the three days leading up to 325 IPOs and then on the dates of those IPOs.

In addition to concluding that the mood of the tweets and an IPO’s performance appears to be related, they also found that:

  • Twitter sentiment during the three-day run-up is usually opposite to how the IPO will unfold. On average, pre-IPO raves on Twitter are associated with price decreases. Conversely, gloomy advance tweets indicate a possible good bet that the price will climb on offering day.

Therefore, write Liew and Wang, advance Twitter sentiment “should be a good proxy for the IPO’s returns on the first trading day.”

  • From open to close on the first trading day, Twitter sentiment mirrors the IPO’s performance, whether positive or negative. While worth noting, says Liew, this contemporaneous information is less likely than pre-IPO Twitter sentiment to give an advantage to investors.

The researchers acquired the data for their working paper from the records of social media analytics firm iSENTIUM, which tracks stock-related tweets and within milliseconds converts the corresponding sentiments into scores (from -100 for most negative to +100 for most positive). These scores then appear on Bloomberg Terminals and other platforms for financial information.

To Liew, establishing the relationship between tweets and IPOs was a particularly intriguing aspect of the study.

“There are close to 290 million monthly active users on Twitter, sending out about a half-billion tweets every day,” he says. “Many of these tweets are opinions and views on the markets, so this unique crowdsourced financial data is growing exponentially. On the negative side, you could say it’s unfiltered noise, and there’s no outside body regulating it. But on the other hand, you could argue that it represents an aggregate view―the wisdom of the crowd. The crowd is telling us something, and it seems worth our while to listen.”

In case anyone doubted Twitter’s impact on Wall Street, Liew mentions the April 2013 phony tweet from the hacked Associated Press account that reported a bombing at the White House, with injuries to President Barack Obama. It led within minutes to a 140-point drop in the Dow Jones Industrial Average. Just as quickly, after the tweet was reported as false, the Dow recovered its losses.

Beyond the stock market―as a predictor of the public mood, a signpost for new strategies, and a provider of a “diverse edge in business intelligence” ― Twitter sentiment can be useful in a range of fields, from health care to entertainment to politics, says Liew.

“We’re still in the early innings,” he adds, “but I think the application of crowdsourced financial information from Twitter is only going to get more exciting as the research community continues to generate new discoveries. The importance of it has definitely been established, and it’s not going away.”

A version of Liew and Wang’s paper, “Twitter Sentiment and IPO Performance: A Cross-Sectional Examination,” can be found at

Wang presented it in March 2015 at a meeting of Data Science MD. In addition, the authors have submitted it for publication in The Journal of Portfolio Management.

Source: Johns Hopkins University Carey Business School