It’s a long-held axiom that news — both good and bad — can move stock prices. Whether it’s news releases issued by the company itself, analysts’ reports, or articles appearing in the consumer press, it’s long been believed that stock prices may change significantly as a result.

The trouble is, nobody had been able to observe that relationship directly, making it impossible to explicitly account for in price forecasts or other sophisticated analyses. That’s changed, thanks to the work of a trio of academic researchers. They developed a news analysis process that not only verifies the influence of news on large changes in stock prices, but also identifies related factors — such as which kinds of news, and for which kinds of firms — will lead to the biggest shifts, or “jumps” in stock prices.

The researchers developed and applied their content analysis methods to an enormous data set of 21 million news articles — including corporate releases and financial press coverage — related to 9,000 publicly traded companies published between 1980 and 2012. (The information is publicly available for any organization or individual wishing to develop their own model.)

“Identifying and forecasting large changes in stock prices is important for many economic and financial decisions,” says Thomas McCurdy, one of the three researchers, and a professor of finance at the University of Toronto’s Rotman School of Management. “If the qualitative content of news about these companies can be analyzed to better identify and anticipate large changes in stock prices, we can make better decisions.

“One could take the results that we’ve had and input them into your favourite forecasting model and have a better understanding or clearer identification of what the main driving factors are of a stock jump,” he adds.

The researchers analyzed the frequency of news for each company and applied natural language processing techniques to the text to identify a positive or negative tone. They next applied a numerical value to that tone, allowing them to quantify the intensity of the tone and how much uncertainty the text communicated.

The single biggest influence on news-related stock price jumps was the frequency of news about a company, the researchers found. Large companies, such as Amazon, Disney and Johnson & Johnson — which represented 80 per cent of the news data set — with high institutional ownership and subject to frequent analyst and media attention, are strongly susceptible to news-related price jumps.

Analyst ratings and news about earnings, capital structure changes, mergers and acquisitions news, and items related to marketing and investor relations, such as conference calls, showed  strong associations with news-related jumps.

This relationship between news and stock prices has only grown stronger in the internet age. The researchers found a five-fold increase in the connection between news and stock prices over the 32-year period they studied, most of it occurring after the year 2000 when the internet was widely adopted. Improvements in the availability and transparency of corporate data and changes in accounting practices and technology have contributed to the rise.

Improving our understanding and modelling of the relationship between news and stock price jumps “can be important for corporate decisions and for the sell-side financial service industry,” said McCurdy. “For retail investors, infrequent but large changes in asset prices can have a very significant effect on the value of their savings and on their portfolio and risk management decisions.”

The research was co-authored by Yoontae Jeon of McMaster University and Xiaofei Zhao, of Georgetown University. Their data set can be found via the online version of their study here.