The jittery, sweaty-palmed fear of losing money in the stock market leaves a signature in Google search data. Upticks in web searches for finance-related words such as debt, stocks and portfolio are good indicators of an impending downturn in the market, a new study shows. People enter such search terms less frequently before market gains, researchers report April 25 in Scientific Reports.
The new analysis is in line with previous work demonstrating that the frequency of some Google searches, such as those related to the flu, can gauge what a large number of people are worrying about.
Some Internet searches are driven by fear, says economist Paul Gao of the University of Notre Dame in Indiana, who was not involved in the study. “Think about it: When you feel good, you don’t search for words like flu or pharmacy.”
Gao’s own research has shown that it’s possible to predict whether the market will bounce back after a bad day by looking at a set of finance-related Google searches that appear to be motivated by fear. He says that the new work emphasizes the need for research dedicated to investigating how the words used in Internet searches reflect how people decide to take particular actions, including buying or selling stock.
To get a sense of whether the search volume of particular words might precede market movements, scientists looked at roughly seven years of Google Trends data showing how often a particular term is queried relative to the total number of searches. The researchers collected data for 98 search terms, including finance-related words such as gold, markets and unemployment and less topical words like kitchen, color and garden. Then the researchers compared the search data with the closing price of the Dow Jones Industrial Average on the first trading day of every week from January 2004 to February 2011. A downturn or uptick in the stock market, they found, was foreshadowed in the preceding days by the frequency of the finance-related terms.
“That was really intriguing to us,” says study coauthor Suzy Moat of University College London. The results are in line with some economists’ belief that people prefer to avoid losing money over reaping profits. If the thought of losing money gives people particular angst, then they may soothe that angst by seeking information, she says.
Moat and colleagues Tobias Preis of the University of Warwick in England and H. Eugene Stanley of Boston University also simulated buying or selling stocks. Working with an imaginary portfolio, they found that buying or selling with the help of Google Trends data could bring in major profits, often far more than what one would reap by buying undervalued stocks and holding them.
That doesn’t mean one should play the market based on search terms alone, however. The researchers observed a trend, but not all words fit the pattern. And if a lot of people were to find out that search terms can be linked to market movements (say by reading Scientific Reports), that signal could become useless or even be manipulated, says Preis.
But the fact that evidence for human decision making can be distilled from an enormous, preexisting data set rather than by conducting surveys or questionnaires is fascinating, says Alessandro Vespignani of Northeastern University in Boston. He and his colleagues are eager to find techniques that use digital traces left by people to predict human behavior. “It’s the meteorology of society,” he says.