Sunday, December 24, 2006

Significant Buying and Selling Days: The Trajectories of Bull and Bear Swings

Well, this isn't the neatest chart I've ever created, but hopefully it will be informative. What we're looking at here is the S&P 500 Index ($SPX) since 2005 plotted against two 20-day moving averages. The first, in green, is the number of significant buying days that occurred within that 20-day period. The second, in red, is the number of significant selling days within the past 20 sessions.

I defined a significant buying day as one in which advancing stocks outnumbered declinining ones on the NYSE by more than 2:1. Similarly, significant selling days represent occasions in which declining stocks have outnumbered advances by more than 2:1. That 2:1 criterion represents more than one standard deviation of strength/weakness in the advance-decline indicator. Fewer than 10% of market days qualify as being significant in their buying or selling. (The total of all significant days--buying and selling--represents roughly 15% of all market occasions).

What the chart illustrates is that there is a common trajectory to bull and bear swings. Let's walk through the phases of bull and bear swings.

As a bull swing matures, the rally becomes more selective and we see fewer significant buying occasions over time (i.e., the green line is dropping, even as the $SPX continues to rise). That is occurring in the present market. The number of significant selling days has not meaningfully expanded over this phase: we're merely seeing less broad-based buying.

The next phase of the market occurs when we see significant selling days interspersed with the muted number of significant buy days. During the early phases of a bull advance, it is common to see the number of significant selling days at zero for a number of days. As the bull swing matures, we see a creep up in the number of significant selling days. By the time the market makes its price peak, we've already had a lift of the significant selling days off of the zero trough.

Once a bear swing takes over the market, the number of significant selling days expands steadily, reaching a peak at or near the market bottom and exceeding the number of significant buying days. Each bear swing we've had in the past several years has had six or more significant selling days out of 20. Note that, overall, significant down days normally occur only about 7% of the time, which translates to 1-2 occasions out of 20. The bear swing represents a cluster of significant selling occasions.

What happens after this cluster is most interesting. We tend to get a flurry of significant buying days, as value-oriented market participants pick up bargains off the recent lows. That leads to a situation in which, temporarily, over a 20-day period, there are *both* a large number of significant buying and selling days. This clustering of significant selling and buying occasions appears to represent the appearance of longer timeframe participants in the market.

From that point, we go back to the initial phase, as the bull swing continues, but with a gradually decreasing number of significant buying days. That is where we're at in the current market. We have not yet seen a meaningful increase in significant selling days.

I've looked over these data going back several decades. The conclusion I am coming to is that it may be worthwhile to separate significant market days from normal, routine ones when analyzing the market. Normal selling days may well scare off traders, but they do not derail a bull swing. Similarly, normal buying days might lead to short-covering, but they do not reverse a bear swing. A "trend" stays in place until longer timeframe participants detect that stocks are trading too far above or below what they deem to be value. Until those participants engage in broad-based buying or selling, a market will tend to continue in its most recent directional movement.

Many trading strategies, over many timeframes, might be developed out of traders' overreactions to normal buying and selling occasions during directional swings. One longer timeframe implication: if we see a market correction that is no more significant in its selling than the last several have been, we can expect the market to resume its upward course. Every correction we've had during 2004-2006 has represented normal selling vis a vis that longer timeframe--not significant selling. An analysis of significant buying and selling weeks and months confirms that perspective.

Dr. Brett's Blog Links and Research Note