Valuable information in early sales proxies: The use of Google search ranks in portfolio optimization

• Author: Josef Zorn,Alexander Kupfer
• Date: 01 January 2019
• DOI: http://doi.org/10.1002/for.2547
• Published date: 01 January 2019

Received: 30 January 2018 Revised: 27 July 2018 Accepted: 3 August 2018

DOI: 10.1002/for.2547

RESEARCH ARTICLE

Valuable information in early sales proxies: The use of

Google search ranks in portfolio optimization

Alexander Kupfer Josef Zorn

Department of Banking and Finance,

University of Innsbruck, Innsbruck,

Austria

Correspondence

Alexander Kupfer,Department of Banking

and Finance, University of Innsbruck,

Universitätsstraße 15, 6020 Innsbruck,

Austria.

Email: alexander.kupfer@uibk.ac.at

Abstract

Weextract information on relative shopping interest from Google search volume

and provide a genuine and economically meaningful approach to directly incor-

porate these data into a portfolio optimization technique. By generating a firm

ranking based on a Google search volume metric, we can predictfuture sales and

thus generate excess returns in a portfolio exercise. The higher the (shopping)

search volume for a firm, the higher we rank the company in the optimization

process. For a sample of firms in the fashion industry, our results demonstrate

that shopping interest exhibits predictive content that can be exploited in a

real-time portfolio strategy yielding robust alphas around 5.5%.

KEYWORDS

asset ranking, entropy pooling, Google search volume, Google trends, Kullback–Leibler divergence,

portfolio optimization

1INTRODUCTION

Timely information often implies an information advan-

tage in finance. Big data are seen as a promising new

source to extract most current information. The use of

freely available big data in finance focuses on the anal-

ysis of news articles (e.g., Smales, 2014; Tetlock, 2007),

social media (e.g., Chen, De, Hu, & Hwang, 2014; Renault,

2017), and Google Trends. For the latter, various streams

of research have emerged: Google search volumeis used as

a proxy for investor attention (e.g., Da, Engelberg, & Gao,

2011; Vlastakis & Markellos, 2012), as a proxy for investor

sentiment (e.g., Da, Engelberg, & Gao, 2015; Preis, Moat,

& Stanley, 2013), and has been linked to BitCoin price

changes (Kristoufek, 2013a). In economics, Google search

volume is mainly applied as a tool to nowcast consump-

tion (e.g., Carrière-Swallow & Labbé, 2013; Choi & Varian,

2012) or to nowcast and forecast unemployment (D'Amuri

& Marcucci, 2017; Pavlicek & Kristoufek, 2015).

Some of these studies further demonstrate that informa-

tion from Google search volume can be used in trading

strategies and portfolio management: Preis et al. (2013),

for instance, apply a trading strategy that links the search

volume of finance-related words with buying/selling the

Dow Jones Industrial Average. For stock-specific applica-

tions, Kristoufek (2013b) shows that linking search vol-

ume to portfolio weights helps to lower portfolio risk and

to outperform an equally weighted portfolio. Similarly,

Bijl, Kringhaug, Molnár, and Sandvik (2016) trading strat-

egy uses firm ticker search volume and buys (sells) stocks

when Google search volume is low (high). Joseph, Win-

toki, and Zhang (2011) classify firms in quintiles based

on Google search volume (high to low). Conversely to Bijl

et al., they apply a long-short portfolio strategy by being

long in firms with high search volume and short in firms

with low search volume. While these studies' strategies

(interestingly) exhibit high abnormal returns, they all lack

an economic justification as it is not clear why changes in

...............................................................................................................................................................

This is an open access article under the terms of the Creative Commons Attribution-NonCommercial-NoDerivsLicense, which permits use and distribution in any medium,

provided the original work is properly cited, the use is non-commercial and no modifications or adaptations are made.

© 2018 The Authors Journal of ForecastingPublished by John Wiley & Sons Ltd

Journal of Forecasting. 2019;38:1–10. wileyonlinelibrary.com/journal/for 1