How Accurately Can Google Predict the Stock Market?

Paul Shearing

Head of Analytics

Industry News

A recent study conducted by UK and US academics has shown that by analysing trends of some search terms you can predict what the stock market will do. The study found that when terms like debt or stocks were searched more regularly than normal a market-wide reduction in share prices would follow. The full report can be read here or a good summary here.

So Why Did This Work?

The actions people take on the stock market are driven by their current feelings and perceptions about what might happen. What people are searching on Google and other search engines gives an excellent indication of those feelings and perceptions. When a lot of people are searching around words like debt and portfolio then it is an indication that people are getting nervous and may follow this by selling shares.

What Are the Flaws?

Over the seven-year period of this study it is clear that there is a correlation between certain terms and the performance of the market. But would the same trend exist over the next seven years? This study was carried out from January 2004 to February 2011, which was a turbulent time for investors. Is it a case that while terms like portfolio correlated to people selling over this period, in a more confident market it would not? The point is that these terms are missing the sentiment behind why people are searching.

What Could Be Possible?

As well as information about how much people are searching on very general keywords, Google also has data for people searching on much deeper search phrases. For example:

  • Facebook Overvalued
  • Facebook Share Price
  • Facebook Dividends
  • Sell Facebook Shares
  • Buy Facebook Shares

Could you use the data Google has on keywords like the above to more accurately predict what is going to happen to share prices? Perhaps when you combine these with more general keywords that might indicate overall market sentiment you could then create a much more powerful tool. It would be interesting to see similar analysis on this more granular keyword data and the correlation to market trends.

There have also been similar studies carried out to see the relationship between activity on Twitter and stock market trends. This data may also be able to give an insight into the missing sentiment of the Google study.

Can This Data Actually Be Used By Investors?

Having read the research your average investor might be getting excited about trying to implement some kind of strategy to replicate the returns the trial received.

However it must be noted that the tools available such as Google Trends do not give up-to-the-minute data. For this to work you need real time data which unfortunately is not readily available.