| dc.description.abstract | We investigate whether search statistics from Google can be used to forecast stock returns over different time horizons. We use daily, weekly and quarterly Google searches, covering the period from 2010 to 2014. The results show a small, positive short-term relationship between daily searches and excess stock returns, a negative relationship between weekly searches and excess returns with subsequent reversal, while quarterly searches are positively related to excess returns without reversal. Next we examine a trading strategy based on our model. The trading strategy shows that there is economical value in including Google search statistics in forecasting models. | |
