Revisiting Seasonality in Overnight and Daytime Returns in the U.S. Equity Markets: Mean-Variance, Sharpe Ratio and Stochastic Dominance Approaches

This paper examines the existence of the day-of-the-week effect in overnight and daytime period returns in a group of broad-index exchange-traded funds (ETFs) that track the major U.S. stock indexes (S&P 500 and NASDAQ 100 indices) over the period from 1996 to 2018. Previous empirical studies suggest that the positive overnight minus daytime mean return spread could be of economic significance. However, empirical evidence is not entirely consistent across studies. To examine this effect, we use various inference procedures: the mean-variance (MV), Sharpe ratio (SR), and stochastic dominance (SD) approaches. The MV and SR results suggest a decrease or even the disappearance of the positive overnight minus daytime mean return spread. The SD results show that overnight periods do not dominate and are not stochastically dominated by daytime period returns, in the sense of first-order SD. These SD findings suggest that no arbitrage opportunities exist in U.S. equity markets and investors could not increase their wealth and expected utilities by switching from any daytime to overnight periods, or vice versa, over weekdays. Overall, the results suggest that information impounding mechanisms have become more efficient in U.S. markets.