Does Google Trends data contain more predictability than price returns?

Damien Challet and Ahmed Bel Hadj Ayed

Abstract

ABSTRACT

Using nonlinear machine-learning methods and a proper backtest procedure, we critically examine the claim that Google Trends (GT) can predict future price returns. We first review the many potential biases that may positively influence backtests with this kind of data, with the choice of keywords being by far the greatest culprit. We then argue that the real question is whether such data contains more predictability than price returns themselves. Our backtest yields a performance of about seventeen basis points per week, which only weakly depends on the kind of data on which predictors are based, ie, either past price returns or GT data, or both.