Making Money in Your Sleep: A Look at Overnight Returns

How many times have you heard the phrase “don’t check the market daily?” It’s something your financial advisor or coach may have told you to help prevent emotionally-driven phone calls and trades. It turns out more than your emotions are at stake: stocks perform better when the market is closed.
Let’s take a look.
Contents
Summary
Equity markets perform better overnight. Holding SPY strictly overnight yields a 4% higher CAGR and experiences 44% less volatility relative to buy-and-hold price return.
Participating in equity markets strictly intraday would have cost you over 2x your starting capital and would leave you underwater even after 27 years of compounding.
Don’t look at your portfolio in the middle of the day. It’s the worst time to be in the market.
If you absolutely must look, look at the opening bell. Theres a 9% higher chance of seeing green first thing in the morning than later in the day.
Return Profiles
Consider the S&P500 price return from Jan 29 1993 to Jan 3 2020. In particular, let’s compare: overnight, intraday and continuous (buy/hold) returns.


Holding the S&P 500 purely intraday underperformed severely, so severely you would have lost your starting investment more than twice over! If you somehow managed to keep your account open you’d still be underwater some 27 years later. Ouch!
Win Rate and Magnitude
Let’s dig a little deeper to see if we can understand why there’s such a large performance discrepancy between overnight and intraday returns. Identifying daily win/lose rate and win/loss magnitude should help.

All approaches yielded a win rate slightly better than 50%. Where performance stands out is with regard to losses.
Holding only intraday experienced 9% more loss occurrences than holding overnight and the average loss was 64.3% larger. To contrast, intraday had 5% fewer wins than holding overnight and the average win was 57.5 larger.
In other words, overnight returns outperform not because they win bigger but because they lose less:

Distribution of Returns
More digging: let’s look at how the returns are distributed. There are 6781 trading days (occurrences) in this dataset. Assuming market returns are normally distributed (hint: they’re not), we would expect to see the following:

What we actually experience is closer to a Laplace distribution – higher peaks and fatter tails relative to a normal distribution:

The width or sharpness of the curve, called kurtosis, is correlated with volatility. The wider or rounder the curve, the greater the volatility.
We also see the curves are slightly off center, favoring positive returns. This makes sense since the win rate is slightly over 50%.
Discussion
So what does all this mean?
Day Trading
My first interpretation of the data is that successful equity day traders are either extremely lucky, found a market inefficiency to exploit or are full of sh*t. Statistically speaking, systematically day trading the broad market from open to close is unprofitable in the long run.
Difference in Returns
The difference between overnight and buy-and-hold CAGR is 30 basis points or a 3.84% underperformance for buy-and-hold. Not that much. There are many ETFs and mutual funds with expense ratios higher than that.
Where the real difference lies is with volatility. Buy and hold was 77.89% more volatile than holding positions overnight.
Leverage
Volatility also matters when applying leverage. If we are comfortable with the volatility of buy-and-hold SPY, we could leverage the overnight approach until we achieve the same level of risk.
Execution of a leveraged overnight strategy could be achieved by:
- purchasing SPY on margin at market close then selling at open
- purchasing SSO (SPY 2x) or UPRO (SPY 3x) ETFs and holding overnight
- trading /ES or MES futures contracts
- options on SPY, /ES or SPX, though you’ll have to deal with theta decay, vega (time value) movements, and other greeks.
Alternatively, one could attempt to hold short positions intraday or use SPUU (SPY -2x) or SPXU (SPY -3x) ETFs.
All leveraged tactics have costs and unique risks which detract from the value proposition of implementing an overnight position strategy.
Practical Application
Trading fees have essentially been eliminated across the board for equity trades. This leaves the costs to: bid/ask spread, low execution quality due to payment for order flow (PFOF), potential tax drag, and of course your time.
Since SPY is the most liquid ETF on the market, bid-ask spread will rarely be more than a penny wide, so this is largely a non-issue.
PFOF should also have minimal impact as the instrument being traded is again the most liquid ETF on the market. Virtually all market participants will have the same bid/ask spread at the same time.
If executing at scale in a [US] taxable account, tax drag will likely offset any total-return outperformance associated with an overnight strategy. Leverage would need to be implemented to overcome this drag, which will further eat into profits and increase the strategy volatility. I haven’t backtested a leveraged implementation of an overnight-only strategy but at a glance there is still much opportunity for total-return and risk-adjusted outperformance even after modest leveraging.
Finally, there is your time investment. This can be in the form of building automation to trade for you, personally attending the market at open and close every single trading day, or hiring someone to do it for you. Depending on how you value your time, this may or not be worth it.
Additional Resources
Private, Custom Backtests
Discover your edge with private, custom backtests for as little as 99 USD. Learn more or contact us for a quote.

Trade Logs
Visit the trade log store and download the data used in this and other backtests.

s1 signal
Trade S&P 500 options more efficiently using the s1 signal. Learn more.
s2 signal
Trade GLD options more efficiently using the s2 signal. Learn more.
s4 signal
Trade Russell 2000 options more efficiently using the s4 signal. Learn more.
mREIT Preferred Share Dashboard
High-yield, low-beta alternatives to cash or treasury bills. Learn more.
Consultations
Schedule a consultation to review your specific scenario and get direct answers to your questions.
January 31, 2020 @ 1:33 am
Hi,
Can you explain What is overnight trading?
January 31, 2020 @ 5:55 am
Certainly! Overnight trading in this context is opening a position at the closing bell and closing the position at the opening bell. Then repeat each trading day.
February 7, 2020 @ 3:46 pm
Yarg, this is super interesting.
Questions:
* Does this hold for QQQ, IWM, etc?
* I’ve heard that Mondays are the worst days for the market. Are there particular days that are better or worse?
February 7, 2020 @ 11:51 pm
Those are great questions! I never considered backtesting other indices or looking at results based on the day of the week. I’ll add those to the list and expand on the existing SPY research. Heck, I may even build an entire portfolio of studies on equities. Stay tuned…
February 8, 2020 @ 1:21 am
I’ve been using a momentum strategy to buy calls in my IRA, and so far it’s been pretty successful. I’m in there every day, so it wouldn’t be much work to buy calls at the close and sell them at the open, but buying leveraged etfs would be a lot cheaper commission wise.
The backtester I use only has close of day data though.
https://tm.cmlviz.com/index.php?share_key=20200207221626_5Pzifua8NwPH3drW
Unfortunately, I don’t have the cajones to hold winners as long as the backtest says to, so I cap my gains.
February 27, 2022 @ 8:19 pm
This strategy won’t work for call buying because of implied volatility (IV crush). It’s possible to lose money on calls overnight even if the underlying is up.
February 9, 2020 @ 8:35 pm
Great work and I’d like to see more! It seems that holding overnight avoids price failure during the day back in Y2K, 08-09, last winter. Wouldn’t that mean that shorting the open and buying the close would be the opposite? One lady only holds the weekends.
January 20, 2021 @ 11:55 am
Spintwig – great work on this article. Do the accretive overnight results hold for more recent times? I.e., if we looked at the last 7 or 10 years only, has this return profile evolved overtime?
The reason I ask is because I noticed the yellow (SPY Buy & Hold) and blue (overnight) lines converged most recently on your “Return Profile” graph. Thus, I am trying to discern whether this phenomenon of outperforming overnight returns still holds OR has the market figured this out since.
January 20, 2021 @ 9:51 pm
Thanks! The return profile has indeed evolved. The trick is discerning between a period of underperformance and an actual loss of edge.
Take AAPL for example. There was a reversal of the overnight outperformance in 2016 per https://spintwig.com/aapl-long-day-trade-equity-backtest/#Discussion
At a glance, it appears there was a similar reversal with SPY in 2016 but subsequent years don’t look as bad as the AAPL example.
February 2, 2022 @ 1:43 am
If this strategy was deployed using $UPRO (triple levered S&P) in theory the results would be the same just 3x correct? My reason for asking is because I know levered ETFs can sometimes get wonky when they rebalance and what not.
February 3, 2022 @ 8:53 pm
Correct.
September 22, 2022 @ 4:42 pm
What backetesting platform did you use? I got much different results using quantconnect.
September 23, 2022 @ 12:26 am
This particular study was done in a spreadsheet. Grab the historical open/close prices of SPY from Yahoo! Finance and measure the close price to the open price the next day.