Study explores how social media can yield signals on financial markets

Social media offers a wealth of signals for understanding financial market behaviour, and the key to leveraging them may lie in distinguishing between how investors feel and what they choose to focus on, according to a new study co-authored by the University of Toronto’s Runjing Lu.

For the research, Lu, an assistant professor of economics in the Faculty of Arts & Science, and co-authors analyzed millions of investor posts on the social media platforms Stocktwits, Twitter and Seeking Alpha between 2013 and 2021. 

“We found that market returns rise prior to high sentiment days, followed by a reversal over the next 20 days – but returns decline prior to high attention days, followed by a continuation of negative returns," said Lu. In other words, when market attention is high, future returns are lower, but after drops in sentiment, returns tend to recover.

The study, published in the Finance, Economics and Banking Research Network (FEB-RN) research paper series, distinguished between how investors feel, or their sentiment, and what investors choose to focus on, or their attention.

“On these platforms, users express both their opinions about how bearish or bullish they are about a stock, which gives us sentiment, and their focus of discussion, the stocks they talk about – which gives us attention,” said Lu.

“We then aggregated these signals across firms and platforms to create separate daily indexes for sentiment and attention that reflect overall market mood and focus.”

This is important because market sentiment and attention have distinct dynamics, and differentiating between the two can help investors more accurately predict market movements.

According to Lu, a trading strategy based on these patterns earns an average excess annual return of 4.6 per cent with a Sharpe ratio – a measure of risk-adjusted return on investment – of 1.2. This represents a solid showing by Wall Street standards.

Another unique feature of the study is its focus on retail – or individual – investors, rather than institutional traders. That focus reflects a trend that has only recently been growing in influence.

“Before COVID, institutional investors dominated the U.S. market,” Lu said. “After the pandemic, with the rise of low-fee brokerage houses, there was an influx of retail investors.

“That’s when everyday people started playing a bigger role in the financial markets, and social media, capturing their sentiment and attention, started to matter more in aggregating information and moving markets.”

Although market-level sentiment and attention are valuable for capturing broad market movements, not all social media information is created equal. An earlier study by Lu and co-authors, published in the Journal of Financial Economics, showed that sentiment from professional investors at the firm-day level is more predictive of next-day returns than sentiment of novices and influencers.

“When you think about who to listen to on social media, it’s the people who have experience and good track records who should be at the fore, not just anyone with an opinion,” Lu cautioned. 

“Social media is not just noise. It’s a real-time reflection of investor psychology and when used carefully, it can offer valuable insights.”