Friday, January 26, 2007

When Markets Take a Body Slam: What Happens Next

Wrestling afficionados know that a body slam has two components: first you lift your opponent into the air, then you slam him onto the mat. On Wednesday and Thursday, we saw a body slam delivered to the equity markets, as a nice gain on Wednesday was more than reversed on Thursday. As I mentioned in my morning comments, it is unusual for a solid rise to turn around and lead to a solid decline; normally there is some rangebound, topping action in between. Thanks to weak housing numbers and rising interest rates, however, selling persisted through Thursday, creating the body slam.

I went all the way back to 1990 (N = 4298 trading days) in the S&P 500 Index (SPY) and measured the day's gains or losses as a percentage of the prior 20 days' average trading range. In all that time, I could only find 15 occasions in which a large rise (one in which the gain of the day was greater than the prior 20 days' trading range) was followed by a large decline (one in which the loss on the day was greater than the prior 20 days' trading range). Five days later, SPY was down by an average of -.43% (5 up, 10 down). That is much weaker than the average five day gain in SPY of .19% (2411 up, 1887 down). Thirteen of those fifteen occasions broke their slam day lows during the following day's trade. It appears that markets that get slammed carry some weakness forward in the near term, though caution should be taken when dealing with such small samples. It is indeed rare for large rises to be followed by large declines.

If we just look at large relative declines (drops in which the daily decline is greater than the prior 20 days' average trading range; N = 360), we can see that those by themselves do not lead to follow-up losses in the near term. Five days after a large relative decline, SPY is up by an average of .39% (211 up, 149 down). Since 2004, we have had 72 occasions of large relative declines; there is modest follow-through weakness in the following two days, with SPY averaging a gain of .02% (30 up, 42 down), but there is no bullish or bearish edge five days out. Interestingly, since 1990, 334 out of the 360 instances (and 64 out of the 72 since 2004) have traded lower than their large decline lows the following day. That suggests very near-term follow-through of weakness.

Finally, I looked at large relative declines as a function of the prior five days' worth of action in SPY. When SPY was up over the previous five days and *then* experienced the large relative decline (N = 155)--which was the case on Thursday, the next two days in SPY averaged a loss of -.16% (67 up, 88 down). When SPY was down over the prior five days and then experienced a large relative decline (N = 205), the next two days in SPY averaged a gain of .37% (124 up, 81 down). It thus appears that we're more likely to get downside follow through on a large relative decline if that decline is correcting prior strength, as is the case at present. Indeed, since 2004, when we've had a large relative decline following five days of strength (N = 42), the next two days in SPY have averaged a loss of -.14% (14 up, 28 down). That is considerably weaker than the average gain of .24% when the large decline follows five days of weakness (N = 30; 16 up, 14 down).

In sum, then, it appears that markets that get slammed tend to follow through with weakness during the following day's trade and, when following a period of strength, over the next two trading days. That will factor into my decision making for Friday's trade.