Sell in May and Go Away: Academic StudyVW Staff
The old, somewhat simplistic, investment strategy, “Sell in May and Go Away” (also known as the “Halloween” strategy), has enjoyed unbroken popularity in the academic literature as well as in actual investment practice. It posits that holding stocks from November through April, and then switching to cash from May through October, provides higher returns and lower risk than a conventional buy-and-hold strategy.
In their seminal study, Bouman and Jacobsen (2002) support the widespread hypothesis that a Halloween strategy offers opportunities for earning abnormal returns. Several more recent studies have confirmed that the effect is still alive and provides profitable investment opportunities (Jacobsen and Zhang, 2012; Andrade et al., 2013, Swinkels and van Vliet, 2012). However, if stock markets are informationally efficient, no such “anomaly” should exist over extended periods of time. As Fama (1970, 1991) and Jensen (1978) emphasize, in a semi-strong efficient market, it should be impossible to profit from publicly available information. And, if such risk-adjusted abnormal returns net of all costs are nevertheless possible, these investment opportunities should diminish quickly as the underlying strategy becomes more transparent.
In fact, it is usually assumed that the first publication of an “anomaly” in the academic literature is of great relevance (Schwert, 2003; Marquering et al., 2006; Jacobsen and Visaltanachoti, 2009). As a result, two issues are important when testing the potential success of a trading strategy: 1) the availability of adequate investment instruments, which makes it possible to effectively implement the strategy, and 2) the date that a trading rule became publicly known and accepted in the investment community. Given that the “Sell in May and Go Away” strategy is easy to implement and very popular among private and institutional investors, there is naturally a question of how it can continue to offer such potential for outperformance. Several recent studies have addressed this phenomenon (Swinkels and van Vliet, 2012; Jacobsen and Zhang, 2012; Zhang and Jacobsen, 2012; Andrade et al., 2013). Most notably, Dichtl and Drobetz (2014) were not able to confirm that the Halloween strategy outperformed a buy-and-hold strategy or any other monthly seasonality-based strategy.
We implement a bootstrap-based simulation framework to test whether the Halloween strategy remains good advice for investors. In contrast to Dichtl and Drobetz (2014), we follow the methodological approach of prior studies, and compare it with only a buy-and-hold benchmark, rather than with all other monthly seasonality-based strategies. In addition to analyzing the time period during which adequate investment vehicles have been available to effectively implement the strategy, we also consider the period when the effect was first documented in a top academic journal (Bouman and Jacobsen, 2002).
We believe our approach will enable us to exploit small datasets in the most efficient possible way. In existing studies, the investment horizon is determined by the length of the available dataset (which results in unrealistically long investment horizons in most cases). We set the investment horizon to one year, because even long-term investors tend to evaluate their portfolios on a yearly basis due myopic loss aversion (Benartzi and Thaler, 1995; Barberis et al., 2001). Furthermore, we implement hypothesis tests from which we can derive conclusions about statistical significance. Our hypothesis tests, which are also based on bootstrap simulations, do not require distributional assumptions for the return and risk measures.
As Jones and Lundstrum (2009) note, in order to ensure backtests are realistic, it is important to use only historical data that was available at the time an investment strategy was implemented. Backtesting further requires that instruments with sufficient market liquidity are available during the entire sample period (e.g., index funds or exchange-traded funds).
Therefore, in contrast to Bouman and Jacobsen’s (2002) study, as well as more recent studies such as Andrade et al. (2013) and Jacobsen and Zhang (2012), we do not use the standard MSCI stock market indices (which begin in 1970 for most developed stock markets).
See Full PDF here: Sell in May and Go Away: Academic Study
Full article via: papers.ssrn