We're suddenly seeing alot of interest in an early version of this 2008 paper over at MoneyScience, so I thought it was definitely worth posting here:

Guy Kaplanski
Bar Ilan University

Haim Levy
Hebrew University of Jerusalem - Jerusalem School of Business Administration

Abstract

In a recently published paper, Edmans, Garc¿a, and Norli (2007) reveal a strong association between results of soccer games and local stock returns. Inspired by their work, we propose a novel approach to exploit this effect on the aggregate international level with the following three unique features: (i) The aggregate effect does not depend on the games results; hence, the effect is an exploitable predictable effect. (ii) The aggregate effect is based on many games; hence, it is very large and highly significant. We find that the average return on the U.S. market over the World Cup's effect period is -2.58%, compared to 1.21% for all-days average returns over the same period length. (iii) Exploiting the aggregate effect is involved with trading in a single index for a relatively long period.

Get the paper from SSRN.