Skip to content

The Overnight Drift (NY Fed): summary and key takeaways

The paper splits the 24-hour S&P 500 e-mini futures day into hourly windows. It finds that one hour, around the European open, carries a large and persistent positive average return. The authors tie it to dealers offloading inventory after sell-offs at the U.S. close.

Boyarchenko, N.; Larsen, L.C.; Whelan, P. · 2023 · The Review of Financial StudiesSummary updated 20 November 2026
Published2023Journal article
Asset focusEquitiesIntraday / overnight
TypeResearchEquity Markets & Premia
Reading time5 minthis summary

The Overnight Drift (NY Fed) at a glance

The question
Why do U.S. equity futures earn unusually high average returns during the opening hours of European markets?
Dataset & period
S&P 500 e-mini futures (most-traded contract), tick-by-tick trades and quotes, hourly returns from mid quotes, January 5, 1998 to December 31, 2020 (23 years). Order-flow and strategy tests focus on 2007 to 2020.
Method
The authors compute average log returns for each hour of the trading day and test them with HAC t-statistics, a block bootstrap and Bonferroni-adjusted p-values. They then proxy dealer inventory with signed futures volume in the last hour before the U.S. close. They sort days by that imbalance and by VIX, run overnight predictive regressions, and use daylight-saving mismatches between the U.S., Japan and Europe as a natural experiment. Finally they test simple long-only futures strategies with and without bid-ask costs.

Key findings of The Overnight Drift (NY Fed)

  • The hour from 2:00 to 3:00 ET, when European markets open, averaged 3.7% annualized, against 3.6% for the whole overnight session and 5.9% for close-to-close.
  • That hour's average return had a t-statistic of 7.1, and after Bonferroni adjustment it is the only clearly significant overnight hour.
  • The drift was positive in 20 out of 23 years and significant at the 10% level in 17 out of 23 years.
  • Negative order imbalances at the U.S. close are followed by strong overnight reversals. Positive imbalances are followed by much smaller reversals, and imbalances near zero produce returns indistinguishable from zero.
  • In volume time, most of the reversal after a sell-off is earned by about 60,000 contracts traded, roughly when Europe opens. Overnight volume is 50 to 100 times lower than U.S. trading hours.
  • A long-only strategy over the drift hour has a pre-cost Sharpe ratio of 1.2. A conditional version that trades only after negative closing imbalances has a Sharpe ratio of 1.80, against 1.3 for the unconditional version over the wider window.

Robustness, as reported in the paper

  • A block bootstrap with 10,000 draws shows the drift hour's point estimates and t-statistics sit above the range for all other hours pooled.
  • Bonferroni adjustment across 24 hourly windows leaves the drift hour as the consistently significant window in the full sample and subsamples.
  • The drift shows up on each weekday and in 9 out 12 months, and appears in both subsamples (January 1998 to December 2010 and January 2011 to December 2020).
  • Daylight-saving tests move the predictability by one hour at the Tokyo open, and shift it later when U.S. and European clocks are out of sync. The paper calls the European test imprecise because it has fewer than 300 days.

Caveats and limits

  • After bid-ask costs, none of the simple strategies is profitable over the full sample. Only the conditional strategy that trades after negative imbalances stays profitable.
  • Signed volume is only a proxy for dealer inventory, and dealer capital is proxied by reported equity value-at-risk. The risk-bearing-capacity explanation for the asymmetry is a conjecture the authors leave for future work.
  • The result rests on one instrument, S&P 500 futures, so it may not carry over to other assets (inferred).
  • The results are mostly in-sample average returns, and the paper does not run a separate out-of-sample forecasting test (inferred).

Items marked (inferred) are our reading of the design, not statements by the authors.

Why it matters

The paper shows that returns on index futures are not spread evenly around the clock. It links one window to a liquidity-provision mechanism, which matters for execution timing and for judging overnight return patterns. It also shows that spread costs can wipe out a calendar effect that looks strong before costs.

Questions about this paper

When does the overnight drift occur?

It occurs in the hour from 2:00 to 3:00 ET, when European exchanges open, with an average of 3.7% annualized.

What explains the overnight drift according to the paper?

The authors argue that dealers who absorb large sell-offs at the U.S. close earn a premium when they offload that inventory to new overseas participants. The reaction is stronger after sell-offs than after rallies.

Is the drift tradable after costs?

Not as a simple long-only strategy. The paper reports that bid-ask spreads remove the profit, and only the strategy conditioned on negative closing imbalances stays profitable.

Where can I read the original paper?

The original is at its publisher: https://doi.org/10.1093/rfs/hhad020. We link to the paper and do not host the PDF.

Follow new quant research

Download MacroFinance market data as CSV or Excel, or pull it by API, and test the ideas in these papers yourself. New summaries are added as we review them.

Create a free account
Summaries are written by MacroFinance from the published papers, in our own words, for information only. They are not investment advice and may omit detail. Papers are not hosted here: each page links to the original at its publisher.