Skip to main content

Day-of-Week & Time-of-Day Effects

Updated Jun 24, 2026 at 2:35pm

Research Draft High 1,233 words

Day-of-week and time-of-day effects are calendar regularities in which average returns, volatility, volume, and liquidity vary systematically with when during the week or the trading session a transaction occurs. The two best-known are the weekend (Monday) effect — the historical tendency for the Friday-close-to-Monday-close return to be weak or negative — and the intraday U-shape — the tendency for volume and volatility to spike near the open and close and slacken at midday. The central tension is that the behavioral day-of-week return anomalies are fragile, contested, and likely arbitraged away, whereas the microstructure time-of-day patterns in volume and volatility are extremely robust and persistent because they reflect the mechanical architecture of the trading day rather than mispricing.

The patterns

Day-of-week return effect. Cross (1973) first documented "non-random movements" across weekdays, and French (1980) formalized the weekend effect: returns on the S&P 500 tended to be negative from Friday's close to Monday's close, not merely diluted by the three-calendar-day gap. Cross reported that over 1953–1970 the S&P composite declined on 60.5% of Mondays (rising only ~39% of the time), versus Fridays, which rose roughly 62% of the time. Proposed (never-settled) explanations include firms timing bad-news disclosures for Friday after the close, settlement-cycle interest effects, dealer inventory behavior, and a weekend deterioration in retail sentiment/mood.

Intraday time-of-day effect. Within the U.S. cash session (9:30–16:00 ET), trading volume, realized volatility, the number of trades, and bid-ask spreads trace a pronounced U-shape — high at the open, lowest around midday, rising again into the close. This stylized fact is documented across equities, index futures (S&P 500, Nasdaq), and FX/commodity futures. The open absorbs overnight information and queued orders; the midday lull reflects information scarcity and lunchtime; the close concentrates portfolio-rebalancing and benchmark-tracking flow.

How it's used in practice

For day-of-week effects, the honest practitioner use is mostly defensive context, not a tradeable signal: knowing the historical Monday tendency exists keeps a trader from over-reading a soft Monday, but the effect is too weak and unstable after costs to trade directly in liquid U.S. equities.

The time-of-day patterns, by contrast, are used operationally every day:

  • Execution / algo scheduling. VWAP and participation algorithms model the U-shaped volume curve so they trade more when liquidity is deep (open/close) and less at midday, minimizing market impact.
  • The closing auction. Because index funds and ETFs benchmark to official closing prices (NAVs), passive flow has migrated to the Market-on-Close (MOC) auction. Closing-auction share of daily volume rose from roughly 3% in 2010 to high-single-digits/~10% by the end of the decade (estimates vary by venue and source). This concentrates price discovery and liquidity into the final minutes.
  • Volatility/risk modeling. Intraday volatility models seasonally adjust for the U-shape before estimating "surprise" volatility; ignoring it makes every open and close look anomalous.
  • Discretionary timing. Many discretionary traders treat the first 30–60 minutes (high volatility, wide spreads, gap resolution) and the closing range as distinct regimes, and the midday "doldrums" as low-conviction.

Adoption, debate & evidence

The split here is critical and often blurred.

Day-of-week return effects are contested and largely faded. A 2024 meta-analysis in Eurasian Economic Review explicitly frames the day-of-the-week effect within the replication crisis, finding that study design — especially the time period sampled — drives whether the effect appears significant. Sullivan, Timmermann & White (2001) argued many calendar effects are artifacts of data-snooping: once you account for the universe of calendar rules tested, statistical significance largely evaporates. Empirically the U.S. weekend effect "immediately declined" after its 1973 publication and has since cycled through reappearance and even reversal (some studies report a positive Monday in later samples). Residual signs survive mainly in small-caps and less-liquid international markets. The fairest reading: a genuine historical regularity, now weak, regime-dependent, and not reliably exploitable net of costs in developed markets — a textbook case of an anomaly decaying once published.

Time-of-day microstructure effects are robust and persistent. The U-shape in volume and volatility is one of the most-replicated stylized facts in market microstructure, holding across decades, asset classes, and venues — because it is generated by the structure of the session (overnight information accumulation, scheduled open/close, benchmark-driven closing flow) rather than by mispricing. It is therefore not an "anomaly" to be arbitraged but a feature to be modeled. The closing-auction migration is well documented and tied causally to passive ownership (e.g., research finding ETF/passive ownership far more associated with auction volume than pre-auction volume).

Strengths & limitations

The time-of-day framework's strength is reliability: it materially improves execution quality and volatility estimation and will not "disappear" because it is structural. Its limitation is that it says when liquidity and volatility live, not which direction price will go — it is an execution and risk lens, not a directional edge.

The day-of-week return effect's weakness is its fragility: results flip with sample period, it is vulnerable to data-snooping, transaction costs swamp the tiny average differentials, and it has decayed in liquid markets. The #1 misuse is treating a historical day-of-week return tendency as a current tradeable rule — e.g., "always buy Monday weakness" — when the meta-analytic and reversal evidence shows the sign itself is unstable. A secondary misuse is conflating the fragile return anomaly with the robust volume/volatility pattern and lending one the other's credibility.

Sources

Dispute flags: Day-of-week return effect is genuinely contested (real vs. data-snooping vs. reversed); exact closing-auction volume-share figures vary materially by source/venue and are quoted as ranges. Time-of-day volume/volatility U-shape is not in serious dispute.