Should The 2% Gap Study Be Adjusted For Volatility?

On Tuesday I posted a study that looked at 2% gaps in the SPY and what the response has been following such a large gap. Afterwards I received an email from a reader who posed the following question:

Rob... a thought on gaps. Perhaps they should be measured in terms of recent ATR volatility rather than an absolute amount like 2%. A 2% gap when ATR_20 is 2% would seem to mean something different than when ATR_20 is 1%.

It’s a good point, and one that has been raised before, so I thought I’d share my answer with you.

The reason I’ve stuck with the absolute number so far is twofold: 1) If you look at the times when the 2% gaps occurred they were all during relatively volatile periods. This somewhat reduces the need to filter by ATR. 2) If I use ATR as the criteria INSTEAD of an absolute 2%, then I will also get a number of gaps that are much smaller included in the study. Is a 1% gap in a non-volatile period the same as a 2% gap in a volatile period? Perhaps from a psychological standpoint it is similar, but if you’re using it to project returns going forward then I’d say it is much different.

Perhaps the best solution would be “2% AND at least an ATR of X”. With only 32 instances to start with I hate to filter too much though.

Unfortunately, I don’t think there is a right way to do it. Hence the reason I typically try and test multiple ideas and take into account not only price movement, but relative volume, breadth and sentiment as well. You get enough things saying “buy” or “sell” and you’ve got a decent chance of being right.