Sell in May Effect: a real deal or a fake one?

Abstract

In an efficient market, information is reflected on market prices, and consequently prices should follow a random walk, and their changes should be unpredictable (Fama, 1965a). The aim of this research is to test the efficient market hypothesis. In particular, this dissertation investigates the weak-form market efficient hypothesis through analyzing the existence of the ”Sell in May and Go Away” calendar anomaly (also known as the Halloween Effect). The data set consists of monthly stock returns for value-weighted indices of 25 international countries. We follow a regression analysis methodology to: test for the existence of the Halloween effect in both developed and emerging markets, test if the Halloween effect is a January effect in disguise, and test if crashes affects the results. The results show that the Sell in May effect exists in 9 countries out of the studied 25 countries, and that the Sell in May effect is not simply a January effect. However, controlling for crashes makes the Sell in May anomaly to disappear. Finally, an evaluation of the performance of both the Halloween trading strategy in comparison with the Buyand-Hold strategy is made. The result show that the former is more profitable in 9 out of the 25 countries even after amending transaction costs.