This is a follow-up to a very interesting post from Brett Steenbarger’s TraderFeed.
Brett divided the S&P 500’s performance (using SPY as a proxy) into changes that occurred from the close to the open (overnight) and those that occurred from the open to the close (daytime) since 2005. I’ve taken a bit wider view and reproduced his results using all SPY data, 1993 to 06/2008, assuming an arbitrary starting value of $1,000.
The three lines above represent the SPY (blue), daytime changes (green), and overnight changes (red). The chart has been logarithmically scaled.
As the chart makes clear, the bullish trend over the last 15 years has been the result of the overnight market, while even during the irrational exuberance of the late 90’s, the daytime market has been timidly flat. Additionally, the overnight market has been the more bear-resistant of the two and exhibited only about half of the volatility. Some stats:
The overnight and daytime markets have been almost completely unrelated. The correlation between overnight and subsequent daytime changes was -7.2%. Between daytime and subsequent overnight changes it was -7.7%. These are very disparate markets.
What’s the point? The point is NOT that we should only be trading the overnight market; there are a number of factors (ex. liquidity and transaction costs) that would need to be considered. The point is that each market presents different opportunities to different types of traders: very active long/short traders might prefer the volatility of the daytime market, while long-only traders might be wary of exiting positions at the close in a bull market.
As I wrote to TraderFeed’s Steenbarger, his post has been stuck in my mind for some time because it’s such a simple yet powerful concept that has been flying completely under my radar. I’ll be keeping this topic on my cutting board and trying to build some sort of workable strategy out of it – more to follow.
Happy trading,
ms
Filed under: Stock Market Mechanics | 3 Comments
Some thoughts -
I have done some poking around in this area as well, and was pretty surprised at the results. This leads me to wonder whether there is some kind of bias for good earnings to reported overnight and bad earnings to be reported during trading hours.
Alternatively, it could also be explained that the overnight markets pretty well move synchronously (althought not in lockstep) with whatever dominant market is currently trading – and foreign markets have blown the US markets away recently due to fleeing foreign interest in the USD.
Regards,
Matt
It would be interesting to see subtract away the market trends to see if the dynamics changed at some point by changes in after hours trading rules.
Perhaps this chart shows that most day traders lose ;^) Meanwhile, the big institutional traders buying and holding big blocks need to use after hours trading by necessity due to sheer volume (or to avoid day traders).
A chart with day/night volume might answer that question.
RE: to peb
Good comment.
I ran a quick graph using the daily after-hours and intraday changes MINUS the daily close-to-close changes. Unfortunately can’t attach that graph to this post, but the basic result is this. The “adjusted” after-hour line is almost completely flat over the 15 years while the adjusted intraday line is consistently decreasing except for the period between about mid-2001 to mid-2003 when it was flat. In a nutshell, I don’t see any type of permanent fundamental change in the observation over the sample.
Happy Trading,
ms