Yesterday, I wrote about some new research showing how tracking search trends on Google can improve forecasts of economic activity. Doing a little more digging after the post, I realized that Google had made the same point back in April.
This paper by Google economists Hal Varian and Hyunyoung Choi demonstrates that the simplest of forecasting models for things like car sales, home sales, and travel were improved by accounting for search trends. Perhaps more importantly, Varian and Choi argue that query data may help economists identify very hard-to-spot turning points.
And it also turns out that this type of data could help make you some money too. Economists Zhi Day and Pengjie Gao of Notre Dame and Joseph Engelberg of UNC-Chapel Hill found that search volume was a good predictor of stock price movements: The more "attention" a stock receives as measured by its search volume, the more likely it is increase in price.
For example, IPO stocks that saw big spikes in Google searches had an average first-day return of 17.3% compared with 10.5% for stocks that didn't get a big influx of attention. This happened even when other things that might grab investor interest like news volume were taken into account.
Search trends also had something to say about price changes of post-IPO stocks:
We ?find that price pressure operates mainly among Russell 3000 stocks with smaller market capitalization ... Russell stocks experiencing large increases in search outperform those experiencing large decreases in search significantly by about 11 bps per week during the ?first two weeks - or 5.7% per year, which is quite sizable for stocks in the Russell index universe.
(The Russell 3000 tracks the 3,000 biggest U.S. public companies.)
The price boost that individual investor attention brings to these stocks is not permanent though: the IPO stocks that got the biggest boost also saw quicker bounce backs. (Because of data limitations, the researchers weren't able to check if popular post-IPO stocks returned to some "fundamental" price. My guess is that, on average, they would.)
The research might help explain why 9 out of the 10 most popular stocks on Google Finance right now are up in price even as the overall market is falling.
Although Da, Gao, and Engelberg warn that their findings don't constitute a profitable trading strategy since they didn't take trading costs into account, it's probably not a bad idea for some enterprising quant funds to pursue this further -- if they haven't already done so.
-- Zubin Jelveh