Opening Gaps in SPY/S&P 500, part 1

Searching on the internet you can find a lot of articles on how to play the opening gap of S&P 500. Today I did my personal twist on this strategy. Over the last two months I’ve been trading a similar strategy, but not exactly the same as the one I’ve tested here.

Here are the details (for long):

  1. If SPY gaps down lower than -0.15% but higher than -0.6%, go long at the opening print/cross. The reason I use -0.6% as maximum is that SPY shows a lot less mean reversion if opening lower. To me that makes sense. Usually there is not that much “news” if SPY opens for example -0.4% down compared to for example 1%. However, if SPY opens more than 1% down it’s a good short, vice versa for long if it opens above 1%.
  2. Target is 0.75 of the gap. If it opens down -0.5%, target is 0.375% higher than the fill price. If target is not reached exit is at the close. No other stops.
  3. Yesterdays close must be lower than 0.25 of this formula: (close-low)/(high-low). The reason I use this is because of SPY’s mean reversion tendencies. It means less fills, but higher average per fill. The tighter I set this criteria, the better average per fill (but obviously less fills).

The test period is from 1. January 2010 until August 2012. In total there is 110 fills and 98 winners. Average per fill is a respectable 0.19%. Here is the equity curve:

Here is the distribution:

Small winners and occasionally a big loser.

The data is adjusted for dividends and collected from Yahoo! and IQFeed. I have tested on both data sets with not much difference. I have used EOD quotes, ie. only open, high, low and close.

I wrote this article on the 20th of September 2012. A perfect day for this strategy. SPY opened down about 0.47% and filled the gap all the way up until yesterdays close.

The strategy seems very robust and yields very good numbers.  Perhaps too good to be true? Yes, it’s probably too good to be true. The reason is that some of the high and low quotes are wrong which boosts the numbers. That’s why I’ll write  second article of this strategy and test it on intraday data from IQFeed. Then we’ll find out if there is a discrepancy in the data sets (which I believe it is).

Any thoughts about this strategy? I know this strategy didn’t perform very well some years ago. However, markets always change and we have to adjust. You won’t find a strategy that last year after year.