In the reading material, it mentions that using the gap return (open_t - close_t-1)/close_t-1 is an improvement over using the daily return (close_t - close_t-1)/close_t-1. And this revision makes the position enter in the open and exit before the close.
I am confused about the whole logic because the gap return only captures the overnight information, which ideally is zero, so how could this feature determine the intraday capital gain? Could anyone help me understand the mechanism behind the gap return? Thank you.
Good question. The key is that the gap return is a mean reversion signal, not a predictor of intraday momentum.
First, overnight returns are not zero. Earnings, news, and global market moves are released outside trading hours, so the gap (open_t minus close_t-1) carries real information. Extreme gaps are often overreactions, thin liquidity, or order imbalances at the open.
The strategy bets that these overreactions partially correct during the day. So we rank stocks cross-sectionally by gap return: buy the big gap-downs at the open, short the big gap-ups, and exit before the close. The gap measures the dislocation, and the intraday session is the holding period where the reversion plays out.
Why is this better than the close-to-close return signal?
Two reasons. The signal is fresher, since we use information right up to today’s open and trade at that same open, instead of trading a day-old signal. And the portfolio is flat overnight, so we avoid gap risk on our own positions.
The trade-off, as the document notes, is roughly double the transaction costs and the practical difficulty of executing exactly at the open.