Can We Predict the Financial Markets Based on Google's Search Queries?

• Author: André A. P. Santos,Marcelo S. Perlin,Martin Pontuschka,João F. Caldeira
• DOI: http://doi.org/10.1002/for.2446
• Date: 01 July 2017
• Published date: 01 July 2017

Journal of Forecasting,J. Forecast. 36, 454–467 (2017)

Published online 17 October 2016 in Wiley Online Library (wileyonlinelibrary.com)DOI: 10.1002/for.2446

Can We Predict the Financial Markets Based on Google’s

Search Queries?

MARCELO S. PERLIN,1JOÃO F. CALDEIRA,2ANDRÉ A. P. SANTOS3AND

MARTIN PONTUSCHKA1

1

Departamento de Administração, Universidade Federal do Rio Grande do Sul, Porto Alegre, Brazil

2

Departamento de Economia, Universidade Federal do Rio Grande do Sul, Porto Alegre, Brazil

3

Departamento de Economia, Universidade Federal de Santa Catarina, Florianópolis, Brazil

ABSTRACT

We look into the interaction of Google’s search queries and several aspects of international equity markets. Using a

novel methodology for selecting words and a vector autoregressive modeling approach, we study whether the search

queries of finance-related words can have an impact on returns, volatility of returns and traded volumein four different

English-speaking countries. We identify several words whose search frequency is associated with changes in the

dependent variables. In particular, we find that increases in search queries including the word stock predict increased

volatility and decreased index returns over the next week. On top of that, we investigatethe performance of a market-

timing strategy based on the search frequency of this word and benchmark it against random words from the Word-Net

database and a naive buy-and-hold strategy. The results of this empirical application are positive and particularly

stronger during the global crisis of 2009. Copyright © 2016 John Wiley & Sons, Ltd.

KEY WORDS investor attention; market efficiency; market microstructure; google trends

INTRODUCTION

The price oscillation of contracts in financial markets is the result of the interaction of a large pool of participants.

Highly capitalized financial institutions and individual investors share their financial views by trading according to

their expectations. An important piece of the puzzle of how markets are organized is related to the way economic

agents behave when faced with different information. From the academic point of view, this is considered a black

box since, for many different reasons, no reliable data can be gathered regarding the behavior of each individual

investor. This leaves us with a large, unexplored gap in the understanding of financial markets’ inner mechanisms,

as we only see the output of the interaction of the different agents in the form of prices and traded volumes, but

never the individual mindset that drives these. In fact, the question of how fast information reaches the participants

and affects their trading decisions has generated one of the pillars of financial theory: the market efficiency theory

(Fama, 1965, 1970).

However, with the advance of technology, we are experiencing a revolution in how social information is collected

and used. The broad collective utilization of Internet search pages such as, Google, Yahoo! and Bing offers rich data

that can be used to better understand systematic effects in the general population as the popularity of the Internet

increases. As a simple example, the frequency of searches for flu symptoms in a particular region of the world can

provide an estimate of the likelihood of a flu outbreak in that area (Dugas et al., 2012). Search frequency data have

been applied to a range of topics: not only the prediction of diseases (Dugas et al., 2012; Ortiz et al., 2011) but also

consumer behavior (Carriere-Swallow and Labbe, 2013; Vosen and Schmidt, 2011), prediction of economic variables

(Choi and Varian, 2012) and others.

Closer to the financial aspect of analyzing Internet search queries, this type of data can be seen as a channel

that allows access to systematic effects impacting market participants. While we cannot see or measure the specific

and individual behaviors of investors, we can at least analyze systematic patterns in social data. For instance, if one

observes an increase in the search frequency of a particular word in period t, this could provide a signal of what the

trading behavior of investors will be in tCk. Additionally, a decrease in the mood of the investors can certainly

impact their trading decisions, which can also impact the frequency of search queries for certain words. Therefore,

the search frequency pattern might also indicate a systematic effect that could be unobservable in any other way.

Internet search queries might also be related to individuals who are looking for news regardingthe financial market,

but with no intention to trade. However, even if their search patterns have no impact on their respective future trading

decisions, it still provides the information that the attention from the population regarding the financial market has

increased, and that has its own potential to affect the expectation of the real traders (Da et al., 2011).

Correspondence to: Marcelo S. Perlin, Escola de Administração (UFRGS), Washington Luís 855, 90010-460, Porto Alegre, Brazil. E-mail:

marcelo.perlin@ufrgs.br

Copyright © 2016 John Wiley & Sons, Ltd