Week 30 Day 3: The Myth of 'Sell in May and Go Away'

Historical data shows the November-April period has produced stronger returns than May-October. But 'stronger' does not mean May-October is negative -- it is just less positive. If you sell in May, you miss the modest gains plus dividends. After transaction costs and taxes, the strategy typically loses money.

The 'Sell in May' data (S&P 500, 1950-2023): Average return November-April: approximately 7.1%. Average return May-October: approximately 3.4%. Both periods are positive. If you sell in May and go to cash (earning approximately 2% money market), your May-October return is approximately 2% instead of 3.4%. You also trigger capital gains taxes on the sale (approximately 15-20% on gains) and forfeit May-October dividends (approximately 0.6%). Net after costs: the 'Sell in May' strategy actually underperforms buy-and-hold in most implementations. The calendar anomaly has weakened significantly since it became widely known (the classic problem: once an anomaly is published, investors trade on it, which eliminates it). Other market timing 'rules' that sound smart but fail in practice: 'Sell before elections.' 'Buy the January effect.' 'Sell when the CAPE exceeds 25.' 'Buy when the yield curve uninverts.' Each has some historical basis but none produces reliable after-cost, after-tax outperformance in real-time implementation. The universal problem: calendar-based and indicator-based rules work in backtests because backtests do not face uncertainty, execution risk, or the psychological challenge of following the rule when it seems wrong.

The 'Sell in May' effect (also called the Halloween indicator) was formally documented by Bouman and Jacobsen (2002) across 37 countries with data spanning 1970-1998. They found statistically significant November-April outperformance in 36 of 37 markets. However, subsequent out-of-sample tests have been less robust. Andrade, Chhaochharia, and Fuerst (2013) found that the effect weakened significantly in developed markets post-publication (consistent with the efficient market response to anomaly publication) but remained somewhat persistent in emerging markets. The economic magnitude: the May-October return is positive but lower, meaning the strategy requires selling to cash, not shorting. After accounting for (1) transaction costs (approximately 0.1% round-trip for mutual funds, higher for individual stocks), (2) tax implications (short-term capital gains in taxable accounts), (3) missed dividends (approximately 50% of annual dividends are paid May-October), and (4) reinvestment risk (money market rates may lag equity returns), Jacobsen and Visaltanachoti (2009) found that the after-cost, after-tax net benefit of the strategy is approximately 0.3-0.5% annually -- a margin so thin that it is easily eliminated by implementation errors (delayed sales, early re-entries, behavioral deviations). For tax-advantaged accounts (IRA, 401k) where taxes are not a consideration, the strategy might add marginal value, but the risk of missing an October rally (which historically accounts for many of the year's best return days) creates an asymmetric risk profile: the most a timer can gain is modest, while the loss from mis-timing is potentially large.