Those interested in market internals, particularly AD breadth, are likely aware of the extreme negative pluralities posted Thursday and Friday of last week (week ending 5-10-06). The negative breadth carnage was across the board, regardless of the index, AMEX, NASDAQ, or NYSE.
The percentage of advances relative to the total number of issues traded on the three exchanges, which will be referred to as the Total Market (TM) breadth, was 21% for Thursday and 22% on Friday. Consecutive days of advancing issues comprising less than 25% of the TM issues traded turns out to be a rather rare event.
The breadth data at my disposal for all three exchanges goes back only to 1978 (AMEX AD numbers being the limiting factor), but I found in the over 7150 trading days in the time span of 1978 until the present, the following statistics:
1) Advancing issues comprising less than the 25% of the TM issues traded for two or more consecutive days has occurred only 30 times over the past 28 years.
2) Of the 30 occurrences, only seven of the events have surfaced since 1990. The majority (17) of such lopsided ratios taking place on two or more consecutive days transpired from 1978 through 1982.
3) Once two or more consecutive days of advancing issues comprising less than 25% of the TM issues traded, an average of 20 trading days elapses before the final intermediate term price bottom is achieved. There is somewhat of a caveat on the number of days elapsing between the negative breadth extremes and the final price bottom, which will be illustrated later.
4) By the time one of the negative breadth plurality takes place, the average severest price decline over the following three months is -4.35%. The price declines ranged from 0% (price bottom coinciding with negative breadth extreme) to -13.33%.
Note the price declines were calculated from the price posted on the final day of the cluster until the ultimate IT low, thus the actual price percentage loss from the highs preceding the cluster were always higher. This study just looks at the price declines following the surfacing of the event.
Below is a spreadsheet listing the dates of each event, the price impact until the lowest price low, the number of days until the price bottom, and the number of consecutive days in the cluster of advances comprising less than 25% of the total issues traded.
In addition, the TM AD McSum value of the day before the first day of the negative breadth extreme, the TM AD McSum value at the final price bottom, and the TM AD MCO on the final day of the negative breadth cluster.
There are five time frames where multiple negative breadth events on consecutive days occurred ("clusters" of clusters). The multiple "cluster of clusters" are highlighted in bold with a shaded background. Within a three month period, the shaded series of consecutive day negative breadth extremes would result in fairly significant and extended price declines; most of the cluster of clusters occurred in the 1978 through 1981. However, there were double digit price declines, see the 1990 event, that were a lone cluster with a price loss of 11.65% over 47 trading days.
There was no particular correlation between the number of consecutive days of a negative breadth debacle and the ensuing price decline. The three largest price declines were kicked off with only two consecutive days of negative breadth plurality.
The larger price declines seem to have taken place when the TM AD McSum were already well in negative territory prior to the cluster events, or at a very high level. However, there is an example of a very modest price decline in 1982 when the TM AD McSum was +1486.
The TM AD MCOs do not provide any consistent clues into the price action following one of the "cluster" events.
So what use is this data for the current puzzle of where prices will go from here? Of the other indicator configurations looked at in this exercise, there appears to be some consistency of the magnitudes of price declines going forward when the TM AD McSum is close to the neutral zone, +/- 250. In the late 70s, there were a couple of 5% to 6% price declines when the TM AD McSum was in the neutral zone, and another nearly 5% decline in 1986 when the TM AD McSum was relatively neutral.
Since the TM AD McSum was at +152 prior to the current cluster event, historical precedent suggests the maximum downside risk, as measured by the SPX, from this point going forward is around -5% over the next three months.
There are undoubtedly numerous other filters that could be used for providing better clues for what to expect in price behavior over the next several weeks based upon the current cluster event, but we'll leave that for future investigations and will react to price action and the internals in the coming days/weeks.