In this post we are going to revisit (check previous post) the catchy market maxim ”sell in May and go away”. After 2 bear markets in the last 3 years and yet another red September, once again, here I am in October, wishing I had sold in May. Let’s simulate the different variations of this seasonal anomaly and see how it is holding up the last 25 years.
The ‘Sell in May’- effect (also referred as the Halloween effect) suggests that stock market returns are effectively higher in the November-April period than those in the May-October period. A popular variation of the saying proposes entering in the market in September (Sell in May and go away come back on Labour Day).
There are various proposed explanations why this anomaly persisted for so long. Among them are the January effect, election cycles, optimism for the New Year and Tax harvesting, the implicit sector-rotation effect of the strategy (avoiding investing the summer months). But there is no convincing argument why this anomaly persists. However, historical data from various global markets suggest that the summer months tend to produce lower returns while being more volatile(many buying opportunities lost). Furthermore August tipically is a rather uneventful month with dull trading activity while September tends to be one of the worst months every year.
Data and methodology
In the following simulations we use the first and last day of the month as well as random entry and exit points in the specified months. For the random portfolios every year the entry and exit date for the selected months is random. In order to make the simulation more realistic, instead of totally liquidating our position we divide it equally in Cash, T-bills, Floating rates and TIPS. Our goal is to retain our purchasing power for the months we are out of the market without assuming much of interest rate risk. The ETFs used for the simulation are:
- SPDR S&P 500 ETF Trust (SPY)
- WisdomTree Floating Rate Treasury Fund (USFR)
- SPDR Bloomberg 1-3 Month T-Bill ETF (BIL)
- iShares 0-5 Year TIPS Bond (STIP)
Although using a 6 month windows and selecting random days in the month fails to produce compelling returns, the old adage holds quite well when we sell near Memorial day or enter in the market near Labour day. In those case we have to mention that we spend time invested in the market. As we see the strategies can not avoid severe drawdowns (2008,2020,2022). Finally the last 3 years this paradoxical anomaly seems to loose momentum.
- Dzhabarov, Constantine S. and Ziemba, William T., Sell in May and Go Away in the Equity Index Futures Markets (January 22, 2016). Available at SSRN: https://ssrn.com/abstract=2721068 or http://dx.doi.org/10.2139/ssrn.2721068
- Dichtl, Hubert and Drobetz, Wolfgang, Sell in May and Go Away: Still Good Advice for Investors? (April 20, 2014). Available at SSRN: https://ssrn.com/abstract=2439280 or http://dx.doi.org/10.2139/ssrn.2439280
- Figueiredo de Castro Junior, Francisco Henrique and Schabek, Tomasz, ‘Sell Not Only in May’. Seasonal Effects in Emerging and Developed Markets (June 24, 2014). Available at SSRN: https://ssrn.com/abstract=2458515 or http://dx.doi.org/10.2139/ssrn.2458515
- Katz, Leslie, ‘Sell in May and Go Away’: Real-Life Listed Securities in the Sherlock Holmes Canon (November 18, 2020). Available at SSRN: https://ssrn.com/abstract=2473500 or http://dx.doi.org/10.2139/ssrn.2473500
- Li, Xuquan and Liu, Ruozhou and Zhang, Zili and Zhao, Xuejun, Buy in May and Sell on St. Leger Day? The Reversal Halloween Effect in QMJ (November 19, 2019). Available at SSRN: https://ssrn.com/abstract=3489865 or http://dx.doi.org/10.2139/ssrn.3489865
- Sakakibara, Shigeki and Yamasaki, Takashi and Okada, Katsuhiko, The Calendar Structure of the Japanese Stock Market: The ‘Sell in May Effect’ Versus the ‘Dekansho‐Bushi Effect’ (June 2013). International Review of Finance, Vol. 13, Issue 2, pp. 161-185, 2013, Available at SSRN: https://ssrn.com/abstract=2262146 or http://dx.doi.org/10.1111/irfi.12003
- Waggle, Doug and Agrrawal, Pankaj, Is the ‘Sell in May and Go Away’ Adage the Result of an Election-Year Effect? (August 1, 2018). Managerial Finance, v 44, 2018; DOI (10.1108/MF-12-2017-0505), Available at SSRN: https://ssrn.com/abstract=3230237
- Jacobsen, Ben and Bouman, Sven, The Halloween Indicator, ‘Sell in May and Go Away’: Another Puzzle (2002). American Economic Review, Vol. 92, No. 5, pp. 1618-1635, December 2002, Available at SSRN: https://ssrn.com/abstract=300700