I’ve received a couple of questions in regards to the Overnight Edge study that I think you may find interesting. As a reminder, the Overnight Edge study demonstrates that points gained in the S&P futures market appear to be much more plentiful during the night sesssion. Again, what I call the night session is after the U.S. close at 15:00 Central time. It was presumed a market edge could be taken advantage of by opening positions during the overnight session or by holding trades after the close.
This overnight edge can appear in individual stocks as well. Below is the example of trading 1 share of Apple (AAPL). No commissions or slippage were deducted. The first equity graph (below) depicts opening a trade at market open (08:30 Central) and exiting the trade at market close (15:00 Central).
You can clearly see the equity graph is very poor – drilling into negative territory. The second chart (below) depicts buying 1 share of Apple at market close (15:00 Central) and exiting at the next day’s open (08:30 Central).
As you can see, there is a dramatic difference between the points gained during the night session vs. the day session. This phenomena is not unqiue to just Apple or SPY. You can find this with other stocks and ETFs. Over at the blog, R Trader, you can see the performance of some ETFs. There you’ll discover that some ETFs perform better during the day while others shine at “night”.
One question I received in regards to the Overnight Edge was how price behaves during the day session. An observant reader pointed out that the day session may actually be productive excluding the final hour of trading. That is, the day session may produce positive returns for long trades if one exited any long trades prior to the last hour of the trading day. The person asking the question thought it was the final hour of trading which destroyed the dailiy returns. Well, this would be easy enough to test, I thought. In order to test this idea I created a simple EasyLanguage strategy on an hourly chart. A given trade was opened at market open and then closed X hours into the day. I then used TradeStation’s EasyLanguage optimization feature to test the exit time from 09:30 to 15:00 Central. Below is a bar graph depicting the exit time vs. the net profit.
As you can clearly see all hours of the day produced a negative return. While it is true holding a trade until the final hour produces the worse results, all hours produce a negative return. This supports the idea that the day session is not producvie for long trades. That is, most of the S&P gains occur during the night session which implies the importance of holding trades overnight or at least, entering trades during the overnight session.
Another reader was wondering, what the night session looked like in regards to maximum adverse excursion. That is, how much did a trade move against you during the night session and how did that compare to the day session. How much “heat” an open trade takes can be an important indicator. My gut instinct tells me the daily price activity produces a higher Maximum Adverse Excursion (MAE) but using my gut for trading decisions is not the best idea. So, I created another simple market study.
The first graph below depicts the MAE for all trades during the day session.
The second graph (below) depicts the MAE for trades during the night session.
Look closely to see that the day session has a cluster of trades which experiences a MAE above $1,000. It appears the day session may hold trades which take on a larger MAE when compared to the overnight session. We can also focus on the losing trades and measure the standard deviation of losing trades. During the night session the standard deviation of losing trades is $277. For the day session the standard deviation is $377. Thus, we have a $100 difference in favor of the overnight trades. I think these results tell us the S&P hold a smaller MAE when compared to the day session. Thus, we can see that trading the night session appears to have two advantages. First, more points gained and second, less heat taken on each trade.
In closing, we can conclude the day session produces negative returns no matter when you exit. Furthermore, the night session trades appear to experience less “heat” than the day hours. All in all, the overnight edge does not apply for all instruments, however for the S&P, there are some strong advantages to the overnight session.
If that is possible, then doesn’t that mean almost every day the SPY will generally experience more red (than green) candlesticks that go higher and higher? Meaning the red candles keep getting higher from each other into the future, instead of green candlesticks. But then when I see the SPY chart, there are still fairly more green candlesticks going higher and higher than red candlesticks going higher and higher.
Remember, the study combines many years of data. Some years may show positive daily return while other won’t. Furthermore, it’ possible to see more green candles then red candles on a given chart yet, the sum of red and green candles produces a negative value. All it takes is for the red candles to subtract more points than points gained from the green. In other words, the red candles are larger.
Nice article! I wrote about this overnight edge in AAPL:
Since last month the edge is gone. It may reappear. This yeat there is a good edge in RTH session vs. the night session.
Thanks Michael. I missed your article about AAPL. I just randomly picked Apple since it was such a popular stock and was very surprised to discover how dramatic the night and day difference was.
I was going to say that this year there is an edge in TLT in the RTH session vs. the night session.
Good article Jeff! I once tried to figure out how XIV perform during the night compared with the day. Like Apple and SPY of course, it shows similar results. What about looking at VIX future contango as a filter? I lack the futures data…
Probably shows that if you want to be long during the day, market open is not the best time to enter.
10:30 is probably the best time to go long. I wonder what the curves will look like if you use 10:30 as the time to go long or exit short.