Brett Steenbarger's "40% of Range" Filter Explored
3 CommentsWhile trying to avoid the seduction of trading the market action after the FOMC statement today, I noticed a Twitter post from Brett Steenbarger. In it, he mentioned his desire to update a November, 2006 blog post from his site entitled, "Do Opening Gaps Tend to Fill?" After reading it, I thought it was worthy of crunching my own numbers.
One of the key findings of his research was that gaps up to 40% of the prior day's range (high to low) have a high probability of filling. Those greater than 40% of the prior day's range only have a 50% probability of filling the same day.
So, I cranked up ol' Otis (my gap fading system) and put him to work to test the gap fade win rates by size of gap relative to the prior day range (note: my system uses Average True Range which incorporates the opening gap into the day's range calculation).
The following shows the results of fading the opening gap of the E-mini S&P 500 futures, 1999 - 2008, targeting gap fill (prior day's close) and using an end of day stop:
Interestingly, the 40% threshold still appears to be the maximum size gap for reliably fading the opening.
Next, I tested all gaps up to 40% of the prior day ATR in size. Here are the results, by year, through yesterday: (note: results are based upon trading 10 contracts)
Win rate dropped a little in 2006 and 2007, but appeared to get back on track (surprisingly) last year and thus far in 2009. Finally, to make it more of a trading rule, I added a stop equal to 30% of the prior day's ATR to see if this simple 40% size filter could actually be worthy of trading:
As expected the win rates drop considerably with the addition of a fixed stop, but still they are quite solid. Profit factors are a little low for my likes, but they are quite good considering the simplicity of the concept. And perhaps most noteworthy of all, the 40% rule appears to be holding up very well during the bear market of the past year and half.
Kudos and many thanks to Brett for this practical nugget that appears to work fairly well in all market conditions. And be sure to check out his site, if you haven't. I always seem to learn something helpful there.
(By the way, the 40% rule would have kept you away from fading today's losing gap fade in the ES. It was 8.75 pts in size and equal to 47% of yesterday's range. )
View Printer Friendly Version
Email Article to Friend
Reader Comments (3)
scott you mentioned that your atr includes the daily calculation of the gap, are you referring to the actual gap at open for that day? i like brett's rule but find gaps
ATR (average true range) by definition includes the opening gap in its calculation. And yes, the size of the gap is measured as simply the difference between the opening price at the beginning of the regular trading hours (i.e. pit session: 9:30 am ET for the e-mini S&P futures) and the closing price of the regular trading hours prior session (4:15 pm ET).
Looks like your comment was cut off - feel free to edit or repost the 2nd part.
The problem I see (perhaps mistakenly but someone please enlighten me as to why) with Brett's rule is that reduced to its basics all it says is that the smaller the gap, the more likely it is to fill. This is not a breakthrough thought. What we take to the bank is not probability but profit expectancy, ie. Profit Factor. To me what enlightening are the statistics Scott has generated showing that % probability holds up quite well around the 50% mark for large gaps. I would have expected much more of a drop-off.