A team of researchers have developed a method which shows that certain patterns in Google searches predict stock market crashes. The team, from Boston University and Warwick Business School, found a way to automatically detect topics people search for on Google preceding stock market falls.
Before declines in the stock exchanges, lead author Chester Curme and colleagues found that searches for politics and business increased when they applied data between 2004 and 2012 to their method.
Their study has been published in the journal Proceedings of the National Academy of Sciences, and is titled “Quantifying the semantics of search behavior before stock market moves.”
A possible warning device?
The authors suggest that their method could be applied to help spot the warning signs in search data before severe stock market fluctuations occur.
Mr. Curme, a Research Fellow at Warwick Business school in England, said:
“Search engines, such as Google, record almost everything we search for. Records of these search queries allow us to learn about how people gather information online before making decisions in the real world. So there’s potential to use these search data to anticipate what large groups of people may do.”
“However, the number of possible things people could search for is huge. So an important challenge is to identify what types of words may be relevant to behaviours of interest.”
In prior studies, Mr. Curme, Suzy Moat and Tobias Preis of Warwick Business School, together with H. Eugene Stanley of Boston University, had already shown that usage data from Wikipedia and Google may contain early warning signs of stock market volatility.
However, those findings depended on the team members selecting an appropriate set of keywords, i.e. finance-related words.
Algorithm automatically raises the red flag
In order to let their algorithm automatically detect patterns in search activity that may be linked to subsequent real world behavior, the researchers quantified the meaning of all the words contained in Wikipedia. This allowed them to categorize words into topics, so that a business topic may contain words such as bank, management and business.
The algorithm identified a wide range of topics, from cricket to architecture to food.
Then, with the use of Google Trends, the team set out to determine how often each week thousands of these words were searched for by online users in the US between 2004 and 2012.
They used these search activity datasets in a simple strategy for the S&P 500, and found that alterations in how frequently users searched for terms relating to politics and business could predict subsequent stock market moves.
The algorithm detected a link between increased “business” and “politics” searches and a subsequent stock market fall.
Historic link between searches and stock market movement
Suzy Moat said:
“By mining these datasets, we were able to identify a historic link between rises in searches for terms for both business and politics, and a subsequent fall in stock market prices.”
“No other topic was linked to returns that were significantly higher than those generated by randomly buying and selling. The finding that political terms were of use in our trading strategies, as well as more obvious financial terms, provides evidence that valuable information may be contained in search engine data for keywords with less obvious semantic connections to events of interest. Our method provides a new approach for identifying such keywords.”
The team believe their findings support their hypothesis that a greater number of searches relating to both business and politics may be a sign of concern regarding the state of the economy, which in turn might lead to less confidence in the value of stocks, resulting in lower-priced transactions.
The research team.
The link is becoming weaker
Tobias Preis said:
“Our results provide evidence of a relationship between the search behavior of Google users and stock market movements. However, our analysis found that the strength of this relationship, using this very simple weekly trading strategy, has diminished in recent years.”
“This potentially reflects the increasing incorporation of Internet data into automated trading strategies, and highlights that more advanced strategies are now needed to fully exploit online data in financial trading.”
Mr. Curme added:
“We believe that follow-up analyses incorporating data at a finer time granularity, or using other types of online data, could shed light on how the relationships we uncover have evolved in time.”
The researchers suggest that their methods might be applied to create predictive models for a range of other events.