Wednesday, January 31, 2007

Market Context: What Happens After a Flat Week of Trading?

The past five trading sessions have been almost perfectly flat in the S&P 500 Index (SPY) on a close to close basis. My previous post examined how we can handicap the odds of moves out of a flat period by observing what other, correlated markets have done during the flat period. In this post, we'll look at how the flat period is situated within the movement of the longer time frame--the context of the flat period--as a tool for determining the odds of upward or downward directional movement. Specifically, we'll look at how the S&P 500 Index moves in the five days following a five-day flat period (i.e., a five-day period in which the index does not rise more than .20% and does not fall more than -.20%).

Going back to 2004 (N = 769 trading days), we have 93 periods of flat performance. Five days after the five-day flat period, SPY averages a gain of .27% (58 up, 35 down). That is modestly stronger than the average five-day change for SPY for the entire sample of .16% (439 up, 330 down).

When both the five and the twenty day periods prior to the flat period were up in price (N = 49), the next five days in SPY have averaged a gain of only .09% (31 up, 18 down). When both the five and twenty day periods prior to the flat period were down in price (N = 16), the next five days in SPY averaged a substantial gain of .80% (12 up, 4 down). We thus see better performance following flatness in a down market than flatness in an up market.

Overall, when the 20 days prior to the flat period were up (N = 67), the five days following the flat period were up by an average of .15% (42 up, 25 down). When the 20 days prior to the flat period were down (N = 26), the next five days in SPY averaged a healthy gain of .59% (16 up, 10 down).

In short, we see during the bull period of 2004-present that flat periods have tended to be followed by strength. The context of the flat period does matter, however, as we've tended to see larger short-term gains following flat periods in down markets than flat periods in up markets. Interestingly, my data suggests that the flat periods in an up market tend to rise in the short run before falling back; the flat periods in a down market tend to fall before rallying. Thus, for instance, 16 of the 26 occasions in which the flat period occurs in a down market take out the prior week's low price. It's thus the trajectory of change--and not just absolute change itself--that is important to the short-term trader. I will have more to say about trajectories in future posts.