Tuesday, December 02, 2008

Historical Patterns and Regimes: A Look at Bearish Momentum Days

One of the chapters of my new book (due out in the first quarter of 2009) explains how I use Excel and historical market data to identify possible short-term trading edges. Unlike some traders, I do not use these historical patterns as mechanical trading signals. Rather, I treat them as hypotheses based on recent market action. If I see markets setting up during the current day in a way that corresponds with the hypotheses from the historical studies, that helps me frame trading ideas for the day.

The recent sharp decline in the stock market has made historical market studies quite difficult, as the current market conditions are quite different from the ones that existed in recent years.

Let's take a simple example. Monday was a very weak day in the market. As those who follow my morning Twitter posts know, I track an indicator called Demand and Supply. Demand is an index of the number of NYSE, NASDAQ, and ASE closing above the upper volatility envelope surrounding a short-term moving average. Supply is an index of the number of stocks closing below the lower volatility envelope. On Monday, Supply exceeded Demand by more than a factor of 10; we had more than 10 times as many issues closing below their envelopes as above.

Such strong bearish momentum has only occurred 16 times since I began collecting these data in October of 2002. To broaden the number of observations for my historical study, I included all occasions since October, 2002 in which Supply exceeded Demand by five times or more (N = 108). What we find is that, after these very bearish momentum days, the S&P 500 Index (SPY) has been up by an average of .31% five days later (64 occasions up, 44 down). The average five-day change for the remainder of the sample was flat (789 up, 763 down).

But not so fast. If we break down those bearish momentum days by when they occurred, different patterns emerge. Since mid-year 2007, we've had 40 days in which Supply has exceeded Demand by five times or more. Five days later, the S&P 500 Index has averaged a loss of -.18% (19 up, 21 down). Interestingly, that five-day average loss followed an average next day gain of .46% (25 up, 15 down).

Prior to mid-year 2007, we had 68 instances of bearish momentum days. The S&P 500 Index averaged a gain of .60% (45 up, 23 down) five days later, with a next day average gain of only .09% (38 up, 30 down).

In short, what we see is evidence of regimes, not universal market patterns. Strong bearish momentum days led to favorable five-day returns from 2002-mid 2007. Since that time, however, strong bearish momentum days have led to one-day snapback rallies, followed by resumed weakness.

When trading historical patterns, it's not only necessary to view market history, but also to look for shifts within that history. Those changing regimes offer fertile hypotheses to traders who can flexibly adapt their thinking.