Skip to content

The Mechanics of the Overnight Drift

Empirical Definition of the Anomaly

The "overnight drift" refers to the phenomenon where U.S. equity returns exhibit large, positive gains during the opening hours of European financial markets1. Specifically, the largest positive returns in the S&P 500 futures market occur systematically between 2:00 and 3:00 Eastern Time (ET), averaging 3.7% on an annualized basis, or 1.48 basis points per day1.

Inventory Risk and Demand for Immediacy

Theoretical market microstructure literature attributes the overnight drift pattern to inventory risk management by market makers1. Based on models of demand for immediacy (such as Grossman and Miller, 1988), market makers absorb order imbalances at the close of the U.S. trading day and demand compensation for bearing overnight inventory risk1.

The structural mechanics of this liquidity premium operate through the following sequence:

  • Selling pressure during regular trading hours results in a negative order imbalance by the end of the U.S. trading session1.
  • Market makers absorb this pressure by becoming net buyers and bearing inventory risk1.
  • These market makers require compensation for bearing this risk, which manifests as positive expected returns when they sell to new market participants arriving overnight1.

Demand Shock Asymmetry

A critical driver of unconditionally positive overnight returns is the asymmetric reaction of prices to demand shocks1.

  • Price reversals following market sell-offs are significantly stronger than the reversals following market rallies1.
  • This demand shock asymmetry gives rise to the unconditionally positive overnight drift returns observed at the European open1.
  • The asymmetry is primarily driven by the time-varying risk-bearing capacity of market makers1.
  • When market makers face value-at-risk (VaR) constraints, positive shocks to volatility—which typically accompany market sell-offs—tighten these constraints1.
  • This tightening magnifies the effective risk aversion of the dealers, requiring higher compensation for providing immediacy during market downturns compared to rallies1.
  • Double sorts on order imbalances and uncertainty confirm that price reversals are larger following days with large end-of-day order imbalances, particularly when those imbalances coincide with periods of elevated uncertainty, such as high VIX levels1.

Clock Time vs. Volume Time

A common structural question regarding this anomaly is why price reversals do not occur immediately upon the opening of overnight Asian markets1. The delay in the price reversal is a function of market liquidity and trading volume1.

  • The close of regular U.S. trading at 16:15 ET marks the only time of the day when trading volumes jump discontinuously downward1.
  • Between 2009 and 2020, trading volumes during regular Asian hours (18:00 – 2:00 ET) were 50 to 100 times lower than volumes during U.S. trading hours1.
  • Due to this exceptionally low liquidity, market makers cannot rapidly offload their accumulated inventory1.
  • When market data is recast from clock time into "volume time" (where time advances by a fixed number of contracts traded), returns increase linearly in signed volume until approximately 60,000 contracts are traded1.
  • This 60,000-contract threshold corresponds to the average number of contracts traded by 3:00 ET, aligning exactly with the opening hours of European markets when liquidity sufficiently increases1.

Elimination of Information Asymmetry Hypotheses

Alternative hypotheses proposing that the overnight drift is caused by asymmetric information or the overnight release of macroeconomic news have been empirically dismissed1.

  • For asymmetric information to explain the overnight drift, news revelation would need to systematically occur between the U.S. market close and the European market open, and this information would need to be unconditionally positive on average1.
  • Empirical testing shows that standard information releases, including macroeconomic data, monetary policy decisions, and corporate earnings announcements, do not explain the overnight drift1.

References


  1. Boyarchenko, N., Larsen, L. C., & Whelan, P. (2022). The Overnight Drift. Federal Reserve Bank of New York Staff Reports, no. 917.