The rules
Buy at the 16:00 close and sell at the next 09:30 open, every trading day, and hold cash during regular hours, with no filter or stop-loss. This is the night interval of Cliff, Cooper and Gulen and the close-to-open return of Kelly and Clark; the studies measure returns over the night and state no position size.
- Universe
- SPY
- Bars
- 1-minute
- Position size
- 97% of equity in SPY each night, cash during the day. Published: none stated; the studies measure returns over the night.
- Rebalance
- Every session, from the close to the next open. Published: the same.
What the research found
Cliff, Cooper and Gulen (working paper; July 2007 draft as summarized by CXO Advisory, revised on SSRN in 2008) used transaction data for 1993-2006: SPY, a fund that tracks the S&P 500, averaged 5.36 basis points (hundredths of a percent) a night against -1.42 by day; costs were not addressed. Kelly and Clark (Journal of Asset Management 2011) found the same pattern in QQQ, a fund that tracks the Nasdaq-100, over 1999-2006. Lou, Polk and Skouras (LSE FMG Discussion Paper 961, 2026, building on their 2019 Journal of Financial Economics paper) put SPY's average quarterly return above Treasury bills over 1997-2023 at 1.84% overnight and 0.21% by day, taking the open as the first half hour's volume-weighted average price. Price Action Lab (2020; SPY, 1993-2020) found 10.2% a year and a Sharpe ratio (return per unit of volatility) of 0.95 before costs, against buy-and-hold's 9.7% and 0.51; with $0.01 a share for commission and slippage (fills worse than quoted) they fell to 3.4% and 0.31. Hu, Pan, Wang and Zhu (Journal of Financial Economics 2022), on S&P 500 index futures, contracts that track the index, from September 1994 to May 2018 (the smaller E-mini contract from September 1997, the standard one before), found 1.99 basis points a day close to open on all days (significant) but 0.69 (not significant) on days without an FOMC, payrolls, ISM or GDP announcement (the Fed's rate decisions and three major economic reports). Boyarchenko, Larsen and Whelan (Review of Financial Studies 2023; E-mini 2004-2020) found close-to-open holding earned 4.62% a year and open-to-close 4.20%; after paying the gap between buying and selling prices, close to open earned 0.38% a year (Sharpe -0.04). Their 2026 note finds that the 2-3 am window, which earned about 3.7% a year in 1998-2020, has averaged near zero since 2021. Falkenstein (2025) finds the overnight edge much weaker since about 2009, with earlier data dominating the long-run averages, and SPY's overnight return negative from July 2022 to August 2023. An overnight S&P 500 fund, NSPY, launched in June 2022, had lost 6.9% by July 2023 while the index rose 22%, and closed in August 2023 (Bloomberg). The published gross edge is about 2 to 5 basis points a night on SPY and S&P 500 futures (derived by Orbital from these figures), against the library's cost of 10 basis points (0.10%) for a buy and a sale.
Results
| Measure | Orbital test | SPY, same days |
|---|---|---|
| Window | Oct 30, 2023 to Oct 30, 2026 | Oct 30, 2023 to Oct 30, 2026 |
| Total return | −24.65% | +88.60% |
| CAGR | −9.00% | +23.54% |
| Sharpe | −1.44 | 1.22 |
| Discounted Sharpe | −5.48 | — |
| Max drawdown | −26.9% | −18.7% |
| Win rate | 48.7% | — |
| Profit factor | 1.34 | — |
| Trades | 753 | — |
Hypothetical: a backtest with modeled costs, not traded with real money.
Hypothetical growth of $10,000
R14 Overnight Hold · −24.6% vs SPY +88.6%
Rebuild it in Orbital
Paste this into Orbital Chat. The Strategy Developer builds the workflow and asks before it runs any backtest.
Build a program on 1-minute bars for SPY, using the session clock, so early closes are handled. Every session, if flat, buy SPY with 97% of equity when the bar before the closing-print minute closes (15:59, or 12:59 on early-close days); it fills at the open of the closing-print minute (16:00, or 13:00 on early-close days). When that closing-print bar closes, sell the whole position; the sale fills at the next session's 09:30 open. Long only; uninvested capital stays in cash. Backtest the last 3 years with $100,000.
Is it real?
- Strategies tested together
- 10
- Beta to SPY
- 0.35
- Alpha vs SPY
- −19.6% a year
- Costs paid
- $65,772, 160% of the gain before costs
- Data
- 1m decisions on 1-minute bars, 754 sessions
- Fills
- Next open at market, sized at the fill
Read this before you trust it
The entry fills at the opening price of the 16:00 minute bar (13:00 on early-close days), not at the official close; in an earlier engine test on another fund, that price was never more than 0.22% from the official close, and the differences averaged out to about zero. The exit is the next session's 09:30 open, as published. SPY stands in for the stocks, indices and futures in some of the studies. On the morning each quarterly dividend comes off SPY's price, the position held overnight is paid that dividend in cash, offsetting the lower open. 97% of equity, leaving room for costs. Two fills every session at 0.05% slippage, about 504 a year.
Sources