Stock Market Predictions and Google Search Data

by Michael on Apr 29, 2013 · 2 comments

Google Logo

Would you be surprised if I told you that Google search volume for various keywords is strongly influenced by real-world events? Probably not.

For example, searches for “who is running for president” spike in January and October of Presidential election years, corresponding to the beginning of the primary season and the runup to Election Day.

But can these sorts of data be used to predict the future? And, if so, is it possible to profit from such predictions? As it turns out, the answers appear to be yes and yes. Or at least that’s what the authors of a recent study would have you believe.

More specifically, researchers at the Warwick Business School and Boston University analyzed the predictive value of 98 search terms that are conceptually related to the stock market. In doing so, they found evidence that query volume for certain terms could potentially be “used in the construction of profitable trading strategies.”

How profitable? Well, if you had traded on the right terms* during the period in question (2004-2011), they show that you could have outperformed a traditional buy-and-hold strategy by twenty-fold. Sounds pretty good, huh?

*Note: I would be remiss if I didn’t point out that it’s also quite possible to underperform the market — perhaps dramatically — if you actively trade stocks based on the wrong terms. 😉

In short, they believe that search data can be used to identify periods of investor concern before people actually act on these concerns, ultimately precipitating a drop in the stock market. By extension, they think that this sort of knowledge can be used to amplify investment returns.

By trading on variation in search volume for the term ‘debt,’ for example, they were able to achieve a total return of 326% from 2004-2011 vs. 16% for buy-and-hold. Of course, they’re backtesting so there’s no telling how well this would work going forward — especially now that the cat is out of the bag.

The big challenge here, of course, is identifying profitable terms in advance vs. after the fact. Would ‘debt‘ work equally well over all time periods? I doubt it. After all, the past few years have seen a huge amount of market-moving drama surrounding things like the debt ceiling debate. And the bulk of the outperformance on this term came during that same time period.

Another big concern here is the potential for false positives. That is, chance correlations between search terms and market performance. For reference, here are the top ten terms based on US search volume:

  • debt
  • color
  • stocks
  • restaurant
  • portfolio
  • inflation
  • housing
  • dow jones
  • revenue
  • economics

While the potential value of most of these terms is obvious, the situation for others is less clear. For example… Color? That’s the second best term in their list. But would you stake your financial future on a trading strategy based on Google search volume for the word color? I sure wouldn’t.

That being said…

I’d be shocked if there weren’t hedge funds out there trying to turn a buck by incorporating this sort of data into their trading models. Who knows? Given enough statistical expertise and computational power, maybe this information really could improve your returns. But for the average investor on the street? Not so much.

Of course, if there is money to be made by trading on Google search behavior, it will be short-lived. After all, once they’re discovered, these sorts of market inefficiencies don’t last very long. But either way, it’s an interesting idea.

Source: Warwick Business School

1 Brick By Brick Investing | Marvin April 30, 2013 at 11:29 pm

This is a very interesting concept. One that could be used as another indicator alongside other useful indicators in their arsenal.

2 Free Money Minute May 1, 2013 at 7:47 am

I really believe the presidential cycle and politics in general have a great influence on the stock market. It doesn’t surprise me that search results also have an effect. We are so spooked as a society these days. I would recommend leaving some money in for the long run and diversifying in other ways as well to make sure you have money when you need it down the road.

Comments on this entry are closed.

Previous post:

Next post: