PROFITABILITY OF TRADING STRATEGIES BASED ON GOOGLE TRENDS

Abstract

This paper seeks to explore the usefulness of data from Google Trends in identifying groups of stocks which may outperform the market. Using Google Trends data as a means for sorting stocks into quintiles weekly, I find that stocks in the lower quintiles have better performances after accounting for the popular Fama French factors. Tests that I have carried out suggest that this could potentially be attributed to the perceived risks by investors on stocks which they have insufficient information on. A back-testing strategy on the lower quintiles modestly outperforms the market after accounting for transaction costs and hindsight bias. However, I find that the results are not stable across time, raising doubts over the long term potential of utilizing Google Trends data for portfolio formation.