The NY Fed’s research staff has a real knack for finding unexpected nuggets that frame US equity market behavior. Back in 2011, for example, they found that essentially 100% of S&P 500 returns from 1994 to 2011 came in the 3 days around Fed meetings. They then updated their findings in 2018 and reported that this “Fed drift” had weakened and now only applies to meetings where the FOMC releases new Summaries of Economic Projections (a least that’s how it worked in 2019).
Last week the Fed staff released a new bit of research with a surprising finding: from 1998 to 2019 “almost 100% of all the US equity premium (was) earned during a 1-hour window” from 2 am to 3am East Coast US time. Here is how they came to that conclusion:
- They looked at hour-by-hour prices for S&P 500 e-mini futures contracts, which started trading in late 1997 and then calculated hourly returns using the most liquid contract.
- The comparison of average hourly returns through a trading day from 1998 to 2019 looks like this, with a notable spike in positive returns visible in the upward pointing blue bar at 2-3 am EST and contrary to what normally distributed hourly returns would look like (red line):
- This “Overnight Drift” is visible in 18 of 20 years in the data, statistically significant in 16 of 22 years, and only negative in 2002 and 2008. It is also not seasonal, appearing in every month and statistically significant in 9 of 12 months.
As for why this “Drift” exists, the authors show that “intraday returns are negatively related to the closing order imbalance (for e-minis) of the preceding day”. They posit that market makers use overseas markets in London/Frankfurt opens to offload positions (long or short) acquired the prior day during the US close. This theory seems to work well, because once Tokyo began trading e-mini futures in 2010, some of the “Overnight Drift” shifted to Asian market opening hours.
So, a neat piece of detective work, but what does it mean for US equity investors?
- American “buy and hold” investors make their returns when they are asleep. Daylight returns average to something close to zero over the long term.
- During stressed markets (2002, 2008, perhaps now), overnight returns after a bad close do not mean revert by the next morning. That makes sense, since there is likely further selling in offshore markets that limits their ability to absorb excess inventory from the NY close.
- As a strategy, US equity day trading must be 100% hedged long/short since the hours of 9:30 am to 4:00 pm do not show the upward bias exhibited by long-run equity returns.
Summing up: as much as the underlying message of this work may be comforting (make money while you sleep), the data shows this does not work in bearish global equity markets. One more reason, we would remind you, to tread lightly just now. We remain on “crash watch” (a down 5% day) to show an investable wash out, even though the Fed’s paper shows why it may not lead to an immediate bounce back if market makers are clogged with sell orders from a sloppy close.
NY Fed Paper (Boyarchenko, Larsen, Whelan, 2020): https://www.newyorkfed.org/medialibrary/media/research/staff_reports/sr917.pdf