How just 10 news photos can predict global daily stock market returns

How just 10 news photos can predict global daily stock market returns

Research shows how investors can use news photos to better predict daily stock market returns.

The more we know about the mood of investors, the better we can predict stock market performance and returns.

While there are already ways to measure investor mood, they are usually via laborious surveys and get outdated quickly.

Now we have a way to get a daily snapshot of investor mood across global markets – developed and emerging – regardless of language.

Earlier this year, University of Missouri researchers devised a way to index daily investor sentiment using news photos in the US.

Research led by RMIT University in Melbourne, Australia has now significantly broadened the scope of this technology, training the algorithm to accurately capture investor mood across global markets – smashing language barriers.

It works by analysing the ‘top lists’ of editorial pictures on stock image website Getty.

Using machine learning, the algorithm produces a daily score based on the types of photos used in global news reports.

Lead author Dr Angel Zhong said investors could use this information to better predict daily stock market returns, based on the mood of investors worldwide.

“You can get a snapshot of global investment mood by looking at the 10 most popular photos, rather than reading hundreds of news articles,” she said.

“This technology could have a huge impact for those wanting to get the feel of the day quickly and accurately.”

Zhong said when people are in a bad mood and facing higher uncertainty, they tend to buy and sell more impulsively and intensively.

“When the photo sentiment measure reflects a bad mood, it predicts a large increase in trading volume,” she said.

“Our study significantly broadens the scope of this tech by investigating how it applies in the international market, across developed and emerging economies.”

Machine learning is already used to scan newspaper text but this method doesn’t overcome language barriers, meaning it’s often only a measure of the developed world.

"If you only look at the text of news articles, you often miss out capturing non-English speaking markets. Analysing images removes that problem,” Zhong said.

“Based on photos in the news, we can predict stock returns in global markets in 37 countries – a healthy cross-section of developed and developing economies.”

Zhong’s study builds on the work of University of Missouri researchers, who trained the algorithm to recognise what a 'good news' and a 'bad news' image looks like, using examples in US media.

Now the algorithm uses those examples, plus more from around the world, to compare future photos and determine a daily market-level investor sentiment index – a daily score for global investment mood.

The Australian team at RMIT and Swinburne University of Technology has trained the algorithm to recognise what constitutes good and bad news coverage from 37 markets, increasing the complexity and global accuracy of the results.

From here, researchers hope to further expand the complexity of the machine learning algorithm by building a measure for each market by using country-specific news photos. The team is seeking strategic partners to accelerate this work.

Photo sentiment and stock returns around the world’, with Angel Zhong, Xiaolu Hu and Mardy Chiah, is published in Finance Research Letters (DOI: 10.1016/


Story: Aeden Ratcliffe


  • Research
  • Business

Related News

aboriginal flag
torres strait flag

Acknowledgement of Country

RMIT University acknowledges the people of the Woi wurrung and Boon wurrung language groups of the eastern Kulin Nation on whose unceded lands we conduct the business of the University. RMIT University respectfully acknowledges their Ancestors and Elders, past and present. RMIT also acknowledges the Traditional Custodians and their Ancestors of the lands and waters across Australia where we conduct our business - Artwork 'Luwaytini' by Mark Cleaver, Palawa.