The overnight/intraday return anomaly is the empirical finding that, when each trading day's total return is split into its overnight (prior close → next open) and intraday (open → same-day close) components, almost the entire long-run gain of US equities has historically accrued overnight, while the intraday component is roughly flat or negative. In other words, the market has tended to rise while it is closed and drift sideways or down while it is open. This is one of the most robust documented regularities in equity data, but it is fundamentally an aggregate, portfolio-level pattern with a contested mechanism and real frictions — not a turnkey single-stock edge.

How it's calculated / formed

Decompose each day's close-to-close return into two non-overlapping windows:

• Overnight return = (today's open − yesterday's close) / yesterday's close. Captures the period the exchange is closed, plus the opening auction.
• Intraday return = (today's close − today's open) / today's open. Captures continuous regular-hours trading plus the closing auction.

Compounded separately over a long sample, the two series diverge dramatically: the overnight series climbs steadily while the intraday series stagnates or declines. The most-cited single illustration is at the individual-stock extreme — Asness, Bryan & Frazzini (Elm Wealth, 2022) note that buying AMC Entertainment at the open and selling at the close every day from early 2019 to late May 2022 would have lost ~99.6% of capital, while holding it only overnight over the same span would have returned roughly +30,000%. That is an illustration of the decomposition's power, not a tradable claim.

How to read it

• A large positive overnight / weak-or-negative intraday split is the canonical signature, present at index level (S&P 500, Russell 2000) and in many individual names.
• The split varies by stock type. This is the "tug of war" of Lou, Polk & Skouras (2019): some stocks earn their return overnight and give part back intraday, others the reverse, and the same is true of factor strategies. They report that high-overnight-return stocks tend to be higher-beta, higher-volatility, higher price-to-book, more "glamorous" names — consistent with a retail/sentiment clientele active around the open.
• The sign of the split is itself somewhat persistent and forecastable: Lou-Polk-Skouras find firm-level continuation within each component plus a slow cross-period reversal, and the smoothed overnight-vs-intraday spread forecasts time-variation in a strategy's close-to-close performance.

How it's used in practice

Few traders trade the raw "buy-the-close, sell-the-open" rule directly (see frictions below). Its practical uses are mostly structural and analytical:

• Style rationale. Because swing and position traders hold overnight, they are structurally exposed to the overnight component that pure intraday/day-traders, who flatten by the close, systematically forgo. This is part of the evidence-based case for the swing style itself — the holding period aligns with where the historical return has lived.
• Execution timing. The finding that opening prices are often elevated and fade in the first ~hour (Cooper, Cliff & Gulen) informs avoiding market-on-open buys; many systematic desks prefer VWAP or closing-auction execution.
• Long/short construction. Academic and quant work builds dollar-neutral portfolios long high-overnight, short low-overnight names; these post very high gross Sharpe ratios but are sensitive to costs.

Standing & evidence

The effect is well-documented and old. French & Roll (1986) first observed that variance accrues differently in trading vs non-trading hours. Cooper, Cliff & Gulen ("Return Differences between Trading and Non-Trading Hours: Like Night and Day," SSRN/2008) decomposed the US equity premium and concluded it was essentially entirely an overnight phenomenon, with daytime returns near zero or negative, holding across NYSE/Nasdaq stocks, indexes, and index futures, and across days/months. Lou, Polk & Skouras (2019, Journal of Financial Economics) tied the pattern to investor heterogeneity — different clienteles dominate the two windows. Hendershott, Livdan & Rösch (2020, JFE, "Asset Pricing: A Tale of Night and Day") found the beta–return relationship is positive overnight but negative intraday, interpreting overnight participants as longer-horizon investors demanding compensation for market risk and intraday participants as more speculative. Bogousslavsky (2021, JFE) maps the cross-section across the two windows. Lachance (2015/2021) found the pattern in nearly every global market.

Proposed mechanisms (genuinely unsettled):

• Overnight risk premium — compensation for bearing close-to-open gap risk when one cannot trade (Hendershott et al.'s reading).
• Overnight news / information flow — earnings and macro releases cluster outside RTH; a 2025 study (arXiv 2507.04481) finds news explains only part of overnight returns.
• Clientele / order-flow timing — retail investors decide in the evening and submit market orders at the open, pushing prices up into shallow opening liquidity, which then mean-reverts intraday (Berkman et al. 2012; Lou-Polk-Skouras). Some authors (Knuteson) controversially attribute part of it to large funds' mechanical rebalancing.

Decay: rolling five-year returns show the index-level effect has weakened markedly since ~2008–2015, after the literature publicized it — a textbook anomaly-decay signature.

Strengths & limitations

• Strength: statistically strong, persistent, internationally pervasive, and decision-relevant for holding-period choice.
• Frictions kill the naive trade. Capturing it requires a round-trip every day; even modest costs erase it. Asness et al. estimate ~1bp round-trip costs cut the long/short ~5%/yr, market impact at ~1% of daily volume adds ~40bp round-trip, and 1990s-style commissions imposed ~100%/yr drag — and real "overnight" ETFs have underperformed the paper index returns by a roughly linear daily drip (cited figures on the order of ~1.7% for an SPY-style and far more for a Russell 2000-style vehicle), confirming the gap is not freely harvestable. (Alpha Architect summarizes the broader "trading costs wipe out the overnight anomaly" result.)
• #1 misuse: treating it as a standalone single-stock timing edge. It is an aggregate/portfolio statistical regularity with a debated cause; the AMC figure is an illustration, not a strategy. It also says nothing about a specific stock on a specific night.

System relevance

This node sits in Market Anomalies & Documented Edges. Its decision-useful link is to the swing style rationale, not to any indicator: see Foundations of Swing Trading (node #1758) and Edges, Anomalies & Seasonality in the swing branch (node #1871) — swing/position traders hold overnight and are therefore structurally positioned to capture the overnight component, which is part of why the style has an evidence base. For the Augustus agent, the operative caveats are: (1) this is a portfolio/aggregate effect, not a per-night per-ticker signal; (2) it is decaying at index level; and (3) it does not imply a literal buy-close/sell-open tactic survives costs. Augustus should treat it as a structural argument for the holding period, with efficacy on any live setup deferred to its own analysis and Cairn's measured record.

Sources

• Cooper, Cliff & Gulen, "Return Differences between Trading and Non-Trading Hours: Like Night and Day" (SSRN abstract 1004081; commonly dated 2008).
• Lou, Polk & Skouras, "A Tug of War: Overnight versus Intraday Expected Returns," Journal of Financial Economics 134(1), 2019, pp. 192–213 (LSE working-paper PDF: personal.lse.ac.uk/polk/research/TugOfWar.pdf).
• Hendershott, Livdan & Rösch, "Asset Pricing: A Tale of Night and Day," Journal of Financial Economics, 2020.
• Bogousslavsky, "The Cross-Section of Intraday and Overnight Returns," Journal of Financial Economics, 2021.
• French & Roll (1986) on trading vs non-trading-hour variance (foundational).
• Asness, Bryan & Frazzini (Elm Wealth), "Night Moves: Is the Overnight Drift the Grandmother of All Market Anomalies?" 2022 — magnitudes, frictions, ETF underperformance, AMC illustration.
• Alpha Architect, "Trading Costs Wipe Out the Overnight Return Anomaly."
• "Does Overnight News Explain Overnight Returns?" arXiv:2507.04481 (2025) — partial news mechanism.