Sell in May and Go Away: A Historically Viable Investment Strategy?

Abstract

This thesis investigates if the efficient market hypothesis holds, despite the pattern of the sell in May effect. Since we cannot study the entire stock market, we have chosen the S&P 500 as a proxy because it is an old and well-tracked index that should be efficiently priced. An OLS regression that compares seasonal returns is used to determine the existence of the effect and the regression shows that there is a statistically significant sell in May effect present in the S&P 500. Another regression is done comparing monthly returns, showing that January has significant positive returns while September has significant negative returns. January is ruled out as the possible cause of the sell in May effect but when controlling for September the difference in seasonal returns becomes insignificant. This indicates that September could be the cause behind the sell in May effect in the S&P 500. Thereafter, we construct two portfolios: one that divests from the market and invests in a T-bill during the summer season, and another that does the same in September. These two lower-risk portfolios, timing the market based on a historical pattern, yield higher returns than the index, therefore contradicting the efficient market hypothesis.