Whether or not you’re a social media user, these platforms are now playing a bigger role in the investment space. In fact, research from Greenwich Associates reveals that nearly 80% of institutional investors use social media regularly at work and 30% of them say information they get on social media has directly influenced an investment recommendation or decision.
But while investors might be scrolling through their social media feeds for new ideas, can they actually use platforms like Twitter and Facebook to anticipate how an investment will perform in the future? Or whether volatility is on the horizon?
New research by University of Washington’s Anthony Sanford delves into the growing and somewhat controversial area of sentiment analysis on social media, looking at whether consumer and investor perceptions can forecast stock returns and volatility jumps.
Using tweets scraped from Twitter, Sanford looks for positive and negative conversations and uses a sentiment index of positive and negative terms to identify how people are talking about a specific stock.
Focusing on data from 2009, the research looks at Apple mainly because the company was the subject of a comparatively large number of Twitter conversations – a proxy, in Sanford’s view, for other large companies like United Airlines, Microsoft, Google, and Samsung. At the same time, 2009 lines up with the release of the iPhone – a tech product that also generated an outsized number of mentions.
As Sanford finds, consumer sentiment gleaned from Twitter posts can not only be used to forecast daily returns but it can also forecast volatility jumps. For insights into the dataset and results, you can download the full paper.
Whether or not this type of analysis and the data it relies on can weather changes that social media companies are making to how its users information is shared remains to be seen. But it does point to the fact that social media is becoming a source of data that could be used by both economists and investors to measure levels of consumer
confidence.