This report was inspired by Quantifiable Edges’ post about the recent large overnight gaps in the stock market and what it portends for future returns (here’s a hint: it’s historically bullish). Specifically, I wanted to test Rob’s assertion that the market is more likely to gap up or down big when in a downtrend.
Previously I’ve shown that (a) bullish markets are realized in the overnight market and bearish ones in the daytime market, and that (b) market volatility is consistently higher in bearish markets than bullish ones. This report will combine these two concepts and wrap it in a pretty bow.
A quick note: in this and future reports I’m going to use a zig-zag concept to define up/downtrends (rather than something too simple like trading above/below a 200-day MA). This sample snapshot is taken from the bear market of the early 2000’s and switches between bull and bear whenever the market moves against a high/low by 10% or more. Bullish periods are marked in green and bears in red.
ANALYSIS
First, let’s look at average daily returns for the ETF SPY in the overnight market from early-1993 to date during up vs downtrends:
Key observations: the overnight market was on average positive in up markets and negative in down markets (no surprise there), but more importantly, the overnight market was much more positive in up markets than it was negative in down markets (this fits with our idea that bull markets are realized in the overnight market). Further, overnight moves were almost 70% more volatile in down than up markets (more on this later).
Next, let’s look at average daily returns in the daytime market during up vs downtrends (SPY, early-1993 to date):
Key observations: the daytime market was about 6-times more negative in down markets than it was positive in up markets (this fits with our idea that bear markets are realized in the daytime market). And like the overnight market, daytime moves were about 50% more volatile in down than up markets.
Lastly, to specifically look at Quantifiable Edges’ observation, let’s look at the average absolute value of daily returns in the overnight market during up vs downtrends. A refresher for the statistically disinclined: absolute value ignores whether a value is positive or negative and just looks at the magnitude (so +1 and -1 both have an absolute value of 1).
Key observation: overnight moves were about 70% larger in down vs up markets.
In summary, these results confirm my assertions that (a) bullish markets play out in the overnight market and bears in the daytime market, and that (b) the markets are generally much more volatile in bearish markets than bullish ones, but also add a new conclusion: Quantifiable Edges is absolutely correct – overnight gaps (as well as daytime moves) are also much larger in bear than bull markets.
How can we apply this to our own trading? Well beyond the obvious warning to be wary of overnight moves (or any moves for that matter) when trading in a downtrend, I’m still 99.99% convinced that there’s a trading system to be had from these discrepancies between the overnight and daytime markets. I just can’t seem to figure out how yet (grrr). More to follow.
Happy Trading,
ms
To stay up to date with what’s happening at the MarketSci Blog, we recommend subscribing to our RSS Feed or Email Feed.
Filed under: Stock Market Mechanics | 7 Comments
One issue you will have when you try to apply this to trading is the inherent lag in defining a bull/bear market (as you probably know, the zig zag indicator peeks). I’m curious how defining bull/bear markets in real time would affect the above results.
Maybe you can use this as a filter – say require SMA of overnight moves to exceed X% in order to take a long trade
RE to Josh: you’re absolutely right. I should have been more clear about why I’m taking this appraoch. The point of the zigzag approach is not to predict the direction of the markets. The point of using the zigzag for these “market mechanics” type of posts is to understand what the market is doing during the bullish and bearish periods. So for instance, a falling market with expanding overnight volatility might be an indication of a bear market (regardless of what the zigzag is saying at that moment). In other words, the logic is reversed…the zigzag isn’t mean to be predictive, it’s meant to help us figure out what is predictive.
Thanks for the comment.
michael
how about buying at end of big down day, & selling rally overnight? eg. today
RE to Dave: I’ve always strayed away from these types of posts b/c Rob Hanna at Quantifiable Edges has always done such a good job at it, but I’ve been thinking about it and I think that I can add some value. Expect a post out this week on it once I get that data together.
michael
Michael, thanks for the explanation. That makes perfect sense and is a very ingenious way to approach the problem.