Thursday, November 09, 2006

Testing Out The Beach Ball Pattern In The Stock Market

We have head-and-shoulders patterns, double tops and bottoms, and flags and pennants. Why not beach ball patterns? You know how beach balls are in the water: You push them down, and they bounce right up.

Wednesday we had a beach ball day, as investors sold stocks on the election news but rallied them higher in the afternoon. What happens after such bouncy occasions?

It turns out that, since 2004 (N = 717 trading days), we've had 16 beach ball days in the S&P 500 Index (SPY), in which the market opened down by more than -.30%, but rallied to close higher than the previous day's close. The next day, the market has been down by an average -.16% (6 up, 10 down). That is notably weaker than the average one-day gain of .03% (390 up, 327 down) for the sample overall.

There's an interesting pattern within the pattern, however. When the market has a beach ball bounce following a five-day period of rising prices (N = 7), as at present, the market has been down an average of -.64% (1 up, 6 down) the next day. That is quite a weak performance.

Conversely, when the beach ball day has follwed a five-day decline (N = 9), SPY has been up by an average of .22% the next day (5 up, 4 down)--stronger than normal.

These are small samples, so must be taken with a grain of salt. Two lessons, however, follow from this little exercise:

1) Subjective impressions in the market aren't always accurate - Before doing any testing, I would have predicted that the beach ball pattern would have been bullish for next-day performance. That seems logical: the market was rejecting lower prices. In fact, if anything, the pattern has been bearish--especially when it follows a period of strength.

2) Context matters - Many chart patterns test out differently depending upon what has happened leading up to the patterns. This is restating something veteran technicians have always known: what's happening on the longer time frame really is important. The beach ball pattern tested out quite differently when it followed strength vs. weakness.

Just about any market pattern you can identify can be tested out. All of my testing is done in Excel with simple open-high-low-close data. Such tests won't always show you a significant edge, but they can be helpful in alerting you to occasions where none exists. In this case, I'm alert to the possibility that the beach ball's bounce may be more of a last gasp for air than a sign of continued buoyancy.