# Overnight vs Intraday Returns Also known as: overnight returns, intraday returns, overnight drift, close-to-open vs open-to-close returns. A Time, Sessions & Seasonality concept (Sessions) in the LuxAlgo Library, with 1 indicator implementation. ## What are Overnight vs Intraday Returns? Overnight vs intraday returns is the split of each day's return into an overnight leg, from the prior close to today's open, and an intraday leg, from the open to the close. The overnight leg is the [opening gap](https://www.luxalgo.com/library/concept/opening-gap/) expressed as a return. In log terms the two legs add up exactly to the close-to-close return, so cumulating each one separately shows where an instrument's return actually accrued: while the market was closed, or while it was trading. The split became a research topic because the answer for US equities was lopsided. In a widely cited working paper, Michael Cliff, Michael Cooper and Huseyin Gulen (2008) found that across US stocks, ETFs and index futures from 1993 to 2006, average overnight returns were strongly positive while returns during trading hours were close to zero and sometimes negative. Henk Berkman, Paul Koch, Laura Tuttle and Ying Jenny Zhang (Journal of Financial and Quantitative Analysis, 2012) traced part of the pattern to attention-driven retail buying near the open, which lifts opening prices that the session then reverses. Dong Lou, Christopher Polk and Spyros Skouras (Journal of Financial Economics, 2019) showed that strategy profits split too: momentum and reversal strategies earn their profits overnight, while the other strategies they studied earn theirs intraday, typically with the opposite sign in the other leg. They attribute the split to different clienteles trading at different times. None of this makes the overnight leg a free lunch. Results depend on the sample period, the market and how the open is defined (an opening auction print, the first trade, or a futures session boundary under [RTH vs ETH](https://www.luxalgo.com/library/concept/rth-vs-eth/) conventions), and the explanations remain debated. Capturing the overnight leg literally means trading at every close and every open, two transactions a day whose costs and slippage can consume the effect. ## How it's calculated Split each day's close-to-close log return into its overnight and intraday legs, then accumulate each leg separately. ``` r_on_t = ln(O_t / C_(t-1)) r_id_t = ln(C_t / O_t) r_t = ln(C_t / C_(t-1)) = r_on_t + r_id_t CumOn_T = Σ over t = 1..T of r_on_t CumId_T = Σ over t = 1..T of r_id_t O_t: session open of day t C_t: session close of day t; C_(t-1) is the prior session's close r_on_t: overnight (close-to-open) log return of day t r_id_t: intraday (open-to-close) log return of day t r_t: close-to-close log return of day t t: day index; T: number of days in the sample CumOn_T, CumId_T: cumulative overnight and intraday log returns over the sample ``` Log returns make the legs add exactly; simple returns compound instead of adding. Ex-dividend price drops land in the overnight leg, so use dividend-adjusted data or the overnight leg will be understated. Define the open and close consistently: official auction prints for stocks, regular-hours boundaries for index futures, and a chosen session boundary for around-the-clock markets such as crypto. ## How traders use it - Diagnosing where an instrument earns: plotting the two cumulative curves side by side shows whether the long-run drift came overnight, intraday or from both, and whether that has changed over time. - Deciding what to hold through the close: a position flattened every evening gives up the overnight leg, so traders weigh that drift against the gap risk measured in a [weekend/overnight volatility profile](https://www.luxalgo.com/library/concept/weekend-overnight-volatility-profile/). - Execution timing: the evidence that opening prices run high in attention-grabbing stocks is one reason some investors avoid market orders at the open for planned purchases. - Strategy decomposition: splitting a backtest's daily results into overnight and intraday components shows when an edge is earned, in the spirit of Lou, Polk and Skouras, and flags strategies that depend on fills at the open or close. ## Overnight vs intraday returns vs related session studies - **Gap-volatility Relation** (https://www.luxalgo.com/library/concept/gap-volatility-relation/): The gap-volatility relation splits variance into overnight and intraday parts to measure risk. This concept applies the same decomposition to returns, asking where the drift accrues rather than where the risk sits. - **Weekend/overnight Volatility Profile** (https://www.luxalgo.com/library/concept/weekend-overnight-volatility-profile/): That profile measures how much movement each closed window contributes. The return split measures the signed return each window delivered; a window can carry heavy variance and little drift, or the reverse. - **RTH vs ETH** (https://www.luxalgo.com/library/concept/rth-vs-eth/): RTH vs ETH defines the session boundaries the split depends on. For index futures, the overnight leg is the electronic session between the regular close and the next regular open. ## FAQ ### Do stocks really make all their gains overnight? In some well-known US samples, essentially yes: average returns during trading hours were near zero while overnight returns carried the market's gain. The finding varies with the period, market and data definitions, so it should be re-measured on the instrument and years you care about rather than assumed. ### Why would returns concentrate overnight? Proposed explanations include compensation for holding positions while the market is closed and cannot be adjusted, attention-driven buying at the open that reverses during the day, and institutional trading during the session that leans against certain strategies. None is settled. ### Can you trade it by buying the close and selling the open? Backtests of that rule have looked attractive in some samples, but it makes two trades every day, relies on auction fills, concentrates gap risk, and depends on a published pattern persisting. After costs the edge shrinks and may not survive. ### Does the split apply to crypto and forex? Only with a chosen boundary. Markets that trade around the clock have no natural close and open, so the split depends on the session picked, such as the US equity hours, and results change with that choice. ## Implementations in the Library - Overnight vs Intraday Returns (LuxAlgo): https://www.luxalgo.com/library/indicator/overnight-vs-intraday-returns/ ## Related concepts - Trading Sessions: https://www.luxalgo.com/library/concept/trading-sessions/ - Intraday Time-of-day Effects: https://www.luxalgo.com/library/concept/intraday-time-of-day-effects/ - Session High/low Statistics: https://www.luxalgo.com/library/concept/session-high-low-statistics/ - Session Open/close Behaviors: https://www.luxalgo.com/library/concept/session-open-close-behaviors/ - RTH vs ETH: https://www.luxalgo.com/library/concept/rth-vs-eth/ - Futures Roll Dates: https://www.luxalgo.com/library/concept/futures-roll-dates/ - Expiration Effects: https://www.luxalgo.com/library/concept/expiration-effects/ --- Source: https://www.luxalgo.com/library/concept/overnight-vs-intraday-returns/ (LuxAlgo Library, the encyclopedia of trading & technical analysis). Free to use with attribution: https://www.luxalgo.com/library/license/