Wednesday, January 9, 2008

How Shall We Bounce?

Markets tend to rally after persistent declines for any number of reasons: program buying, key technical support reached, positive news flow, short covering, bargain hunting -- the list goes on. In any event, market direction is a process and not a straight trip; bounces are inevitable and today was one such day.

Last November, I did a series of articles titled along these lines "NASDAQ 100 Down X Days > Y%: What Happens Next?" Rather than posting an endless series this go around, I thought I would post a more comprehensive study for future reference.

The tables below present average five-day returns after the markets have consecutively declined a given number of days (columns) by increasing levels of cumulative percentage loss (rows) for both the NASDAQ 100 (NDX/QQQQ) and S&P 500 (GSPC/SPY). Over the ten-year study period (n=2,519 days), the baseline average five-day return was 0.24% and 0.11% for the NDX and GSPC, respectively. Importantly, a down-day is defined here as one where average price is lower than the prior, with average price calculated as [(Open + High + Low + Close)/4].




The intersected yellow colored cells represent the state of the respective indices as of yesterday's close (1/8/08). In the last ten years, the combination of duration and level of such pullbacks has been quite rare indeed! While the NASDAQ100 is flat on average five-days later (n=3 first instances only), the S&P 500 typically shows strong gains exceeding 3% (n=6).

This is apparently somewhat typical. Note generally the differences in subsequent five-day performance behavior between the two indices. As perhaps best seen in the topographic charts, while both indices perform best after short, sharp declines, although the S&P 500 tended to outperform the baseline across all combinations, on average the NASDAQ 100 tended to exhibit subsequent downward momentum after a tipping point of seven or eight days of consecutive declines.

My initial thought is that this behavioural difference relates to the NASDAQ 100's greater historic industry concentration, which may make it more susceptible to durable trending patterns when strong fundamental/structural rotations occur. Second, a deeper look at the timing of these few extended pullbacks may likely show that many occurred during the 2000-2002 post-bubble technology correction on valuation. These are obviously related thoughts.

Assuming these hypotheses are correct, I don't have reason to believe that either situation applies currently and would therefore argue that the index should more equitably participate in any general market bounce we may see in the days ahead.

4 comments:

Jeff Pietsch CFA, Esq said...

To the reader asking how the charts where made, they were created using Microsoft Excel 2007 based on the cumulative instances tables. Also, yes the post was up, then down briefly upon posting to get the formatting correct and, as is often the case.

Cumulative instances means it is the average of all days meeting the criteria, including those that fall within the same general decline, looking at the 5-day result forward from the 'first-instance' and the subsequent days within the delcine as well.

Jeff Pietsch CFA, Esq said...

Another thought, all results are a priori, so a three day pull back that turns into a four day pull back is contained in both results. Also, remember these are only averages!

Mike said...

thanks jeff

-guy who was asking

Jeff Pietsch CFA, Esq said...

Thanks Mike, everynow and then (OK, more often than not) Google Blogger flakes out as it did in this case, deleting your posts when I had tried to publish them!