A higher frequency of positive overnight returns followed by negative trading day reversals during a month suggests a more intense daily tug of war between opposing investor clienteles, who are likely composed of noise traders overnight and arbitrageurs during the day. We show that a more intense daily tug of war predicts higher future returns in the cross section. Additional tests support the conclusion that, in a more intense tug of war, daytime arbitrageurs are more likely to discount the possibility that positive news arrives overnight and thus overcorrect the persistent upward overnight price pressure.

Miller [1977. Risk, uncertainty, and divergence of opinion. Journal of Finance 32, 1151–1168] hypothesizes that prices of stocks subject to high differences of opinion and short-sales constraints are biased upward. We expect earnings announcements to reduce differences of opinion among investors, and consequently, these announcements should reduce overvaluation. Using five distinct proxies for differences of opinion, we find that high differences of opinion stocks earn significantly lower returns around earnings announcements than low differences of opinion stocks. In addition, the returns on high differences of opinion stocks are more negative within the subsample of stocks that are most difficult for investors to sell short. These results are robust when we control for the size effect and the market-to-book effect and when we examine alternative explanations such as financial leverage, earnings announcement premium, post-earnings announcement drift, return momentum, and potential biases in analysts’ forecasts. Also consistent with Miller's theory, we find that stocks subject to high differences of opinion and more binding short-sales constraints have a price run-up just prior to earnings announcements that is followed by an even larger decline after the announcements.