[this post came from Percy C, a trader, a scientist, and a highly rated presenter at the recent SpikeTrade Reunion – Alex]

Let us take a look at the violation of the upper Bollinger Band (20, 2) in SPY last week. As folks undoubtedly know, Bollinger Bands represent a demarcation of a (typically) ~2 standard deviation range around a (typically) 20-period simple moving average. There are a variety of uses of the bands: some use them simply as targets, others use them to initiate positions in the direction of the excursion, and still others use them as points to initiate countertrend trades. You can find any number of sites and books that comment on the bands, including Bollinger’s own site, http://www.bollingerbands.com/

Let us compare the historical performance of SPY in the weeks subsequent to a violation of the upper Bollinger Band (20,2). I pulled the 2000 – 2010 data from a reliable vendor and the 1994 – 2000 data from Yahoo! (buyer beware!), and calculated the bands, identified excursions of the upper band, and looked at price changes subsequent to that time. These data are summarized in the table below and broken down into the entire time period tested (January 1994 – last week) and the past decade (January 2000 – last week). You can see the average and median percentage price changes 1 – 55 weeks after a violation. I’ve also provided the maximum and minimum prices changes, to give folks a sense for the best and worst cases, depending on whether or not you like to trade with the excursion or against it.

As you can see, the average % changes are nothing to write home about. I was also impressed that although there have been some nice bullish moves in the 8 week period after the excursions, the analysis of maximum and minimum changes seems to indicate that there is some skewing to downside risk. The examples in the past three years confirm that: four violations in the April – May 2007 time period (when SPY was ~150), one violation in August 2009 (SPY ~101), two violations right before the April 2010 top (SPY 120 – 122), and two violations in the past five weeks. Just scrolling through the numbers, it seems that the excursions occur in all three places that you might expect: (1) in the middle of strong trends, (2) right before periods of consolidation, and (3) right before major topping patterns (e.g. March 2003, May 2007). I didn’t see an obvious edge, but it might be possible to construct a system that would deliver positive returns. [Note also the usual disclaimer: these are my personal observations, and you should personally verify all data before considering any trades.]

Best wishes for successful trading!

Percy C