Overnight vs Intraday Returns

Every trading day is really two returns stitched together. The overnight return is the jump from yesterday's close to today's open — the part that happens while the market is shut. The intraday return is the drift from today's open to today's close, during regular trading hours. Add the two (compounded) and you get the full day. Splitting them apart reveals one of the strangest regularities in equity data.

The finding: for US stocks and the index, essentially all of the long-run positive return has arrived overnight, while the intraday session has, over long stretches, been flat or even slightly negative. If you had only ever held from the close to the next open — never during the day — you'd have captured most of the market's gains. Hold only during the day and you'd have gone almost nowhere.

The decomposition

Write the total gross return over a period as the product of the overnight and intraday pieces:

$$1 + R_{\text{total}} = \big(1 + R_{\text{overnight}}\big)\big(1 + R_{\text{intraday}}\big).$$

Here $R_{\text{overnight}}$ compounds all the close-to-open moves and $R_{\text{intraday}}$ compounds all the open-to-close moves. Because these two components can be tracked separately over years, you can plot how a dollar would have grown holding only one session at a time.

Worked example: splitting a year

Suppose over one year the index returned +10% total. You separately compound the two sessions and find the overnight component was +12% and the intraday component was −1.8%. Check that they multiply back:

$$(1 + 0.12)(1 - 0.018) = 1.12 \times 0.982 = 1.099,$$

which is +9.9%, matching the +10% (rounding aside). So a +12% overnight tailwind more than accounted for the full year, and the daytime session actually subtracted about 1.8%. An investor who mechanically bought at each close and sold at each open would have earned the 12%; one who bought each open and sold each close would have lost money — over a year the market rose.

Why does it happen?

No single cause is settled, but the leading stories are microstructure, not magic:

• Overnight risk premium. Holding through the close means bearing the risk of gapping on overnight news with no ability to trade out; you may be paid for that.
• Order-flow and demand pressure. Retail and ETF buying concentrated near the open and close, and dealer inventory effects, push prices around predictably.
• Earnings and news timing. Most earnings and macro releases land outside regular hours, so their price impact registers as an overnight move.

Why you (probably) can't just trade it

The pattern looks like free money — buy every close, sell every open — but the frictions are brutal, which is precisely why the inefficiency persists.

• You cross the spread twice a day. Trading at the open and again at the close means paying the bid-ask spread and market impact 250 times a year. Those Transaction Costs can easily exceed the overnight edge, especially for smaller stocks where the effect looks largest.
• Open and close are the worst times to trade. Auctions and opening prints are volatile and information-rich; the price you actually get is often far from the "close" printed in the data.
• The edge is thin per name and concentrated in illiquid stocks, so scaling it up runs straight into capacity limits.

The overnight-versus-intraday split is a clean lesson in how much return can hide inside the timing of a position, not just its direction — a cousin of calendar Seasonality Effects and a reminder that a raw pattern and a tradable strategy are two very different things.