All of the S&P 500’s recent returns occurred during the overnight sessions. Whenever this effect is especially pronounced (as it has been since April 2020), it makes the rounds as if it were a brand new discovery. But it has actually been around for at least as long as the major S&P 500 ETF, SPY, has existed.

Another well publicized finding (conveniently ignored by many articles about the effect) is that running a dumb overnight strategy (buy SPY at each close; sell at each open) erodes the entire overnight edge through slippage and transaction costs.

In this post, I will dig into some hypotheses on reasons for the effect and will use my findings to go through the steps of constructing a simple, viable overnight trading strategy. (I have tried this exercise before, and you can read about my experiences with another ultra-short-term strategy “Fading The Opening Gap,” which includes a section on my general skepticism regarding day-trading.)

The other side of the coin

The enormous overnight outperformance during strong market rallies (e.g. the late 1990’s in the chart above, or the post March period in 2020, charts below) stands in contrast to periods of problematic underperformance. 
For example, in March 2020, the entire drawdown happened during the overnight sessions; in mid-2015 to mid-2016, over a turbulent one-year period, the bulk of gains happened during market hours, while at-night returns went through a steady decline; and in the bear markets of 2000-2003 and 2008/2009, overnight sessions saw major underperformance.
In contrast, “Volmageddon,” in the beginning of 2018, and the market losses towards the end of that year are unnoticeable when looking at overnight returns only.

From 2010 to 2016, the overall return is not notably different between night and day periods. But, after the troubled market periods in 2010 and 2011, during the next bull run from 2012 to 2015, the overnight performance was notably smoother, with all major drawdowns materializing during market hours.
Lower volatility causes a much higher Sharpe Ratio for overnight returns, making a leveraged implementation possible using cost efficient futures. To turn this into a viable strategy, however, it would be key to find a way to filter out periods of major drawdowns.

“If you were to ignore March 2020, then the overnight returns of the S&P 500 from 12/31/2017 to 12/31/2020 would be 22.9% annualized and have a Sharpe ratio of 2.1.
(With March, it drops to 14.5% annualized and a Sharpe ratio of 0.91).
In comparison, intraday returns would have a return of -0.4% annualized.” Corey Hoffstein

Since 2017, all of the index returns have once again come from the overnight session — notwithstanding the massive overnight drawdowns during the March 2020 crash, where, amazingly, the day sessions went through that period virtually unscathed.

Reasons for the overnight effect

Low Liquidity
A lot of the academic literature on the subject fails to find a convincing explanation for the persistence of such an obvious effect. The basic theory, as presented in a thorough article in the New York Times, looks rather thin to me:

“Because relatively few people actually trade after the market closes, orders tend to build up overnight, and in a rising market, that will produce an upward price surge when the market opens. But during extended declines, overnight sell orders may cause prices to plummet when the market opens.” New York Times

Just why does that tell us anything more than that overnight gains or losses mostly occur all at once at the market open — which would actually imply that these returns materialize during market hours, wouldn’t it?

Overnight Risk Premium
A risk-based explanation makes more sense in light of the pertinent question: Why haven’t investors used a strategy to exploit this pattern, and thereby neutralized it? A large amount of stock-market risk (e.g. economic data and earnings releases) happens outside market hours, and overnight returns are thus a compensation for taking on that risk.
Conversely, risk-averse traders are inclined to close positions by market close and re-enter during the day.

Structural Liquidity Flows
Looking closer at low overnight liquidity and its effects on forced option market maker positioning yields some additional insights, which can be used to filter out likely weak periods for a viable overnight strategy, as well as lead to a logical timing of entries and exits. It also gives confidence in a likely persistence of the overnight effect in the future, because an edge that is caused by structural liquidity flows is less likely to be arbitraged away, even once it becomes widely known.
In fact, the growing size of the options market in comparison to falling liquidity in equities is making these influences even stronger.

The main culprits causing liquidity flows are Vanna and Charm — two option greeks — because they influence options market maker hedging exposure.

Why do options cause liquidity flows?
Vanna describes the influence of a change in implied volatility on an option’s delta (the option price’s rate of change compared to an underlying). Usually dealers are short puts (and/or long calls) and therefore are long Vanna. 
Contango: 75% of the time volatility markets are in contango (longer-dated VIX futures are naturally more expensive than those closer to expiration, because they can be used as a crash hedge, which is not free), causing the implied volatility of an option to continually fall over time — all else being equal. This enables dealers to buy back S&P 500 futures that they use to hedge their market risk, thereby supporting stock prices.  
Charm or Delta Bleed: the time to expiration influences option price sensitivity to changes in the underlying. Options need to be hedged a little less every day.

Due to overnight liquidity constraints, these flows materialize during predictable time windows. When European markets open in the morning, enough liquidity becomes available for large market participants to be able to adjust their hedging positions. As this is well known, a certain amount of front-running by smaller traders takes place before that. Later in the day, the opening of US markets provides a similar window of increased liquidity.

“Vanna/Charm hedging happens on the close and in the morning. Vol down every day means buying in indices overnight. Vol up every day means selling overnight… Vanna/Charm flows are hedged by banks & naive primary dealers on the open of US trading & into close. Algos & European traders front run these flows starting around 1am CST to sell on US Open. I believe the excess risk adjusted returns are 100% a result of captive Vanna/Charm flows…” Cem Karsan

A paper published by the New York FED supports this empirically, highlighting the overwhelming importance of that first window of liquidity: “We show that nearly 100 percent of the U.S. equity premium is earned over a window around the opening hours of European markets when U.S. cash markets are closed.“

Vanna Nights — creating an overnight trading strategy

We can now take these findings to create an optimized S&P 500 overnight strategy. It uses filters to avoid trading at times when the overnight effect is likely to be negative, or when the edge disappears, and utilizes optimal instruments and leverage rather than buying the SPY ETF for each trade.

Please register for free to continue here with part 2.

Good luck with your investments, and thank you for reading!

David

Browse Recent Blog Posts

If you like what you are reading, please consider subscribing – thank you! Your support is greatly appreciated.