Saturday, February 21, 2009

When Markets Are Broadly Weak

I noticed on Friday that we had an explosion of stocks making new 20-day lows across the NYSE, NASDAQ, and ASE. New 20-day lows hit 4129, a level we've seen only 11 times since late 2002.

Interestingly, when we've had more than 4000 new 20-day lows, there has not been an intermediate-term bullish bias. Indeed, 30 days later, the S&P 500 Index (SPY) has been up 5 times and down 6, for an average loss of 3.0%.

A more interesting observation, however, is that--of the 11 occurrences of 4000 or more 20-day lows--all but one (5/10/04) has occurred since the second half of 2007. During the bull market, pullbacks led to an excess of new 20-day lows, but not such broad weakness. When markets have been broadly weak, it's been an indication of bearish trend conditions and we haven't seen a bullish edge going forward.

An interesting topic for research is whether extremely weak or strong markets differ qualitatively from normally weak or strong ones. It may well be that rising or falling tides that lift or drop all boats show greater odds of continuation, whereas less broad moves are more likely to encounter reversal.

Note: I post new 20-day highs and lows each morning before the market open via Twitter (free subscription via RSS).


Tony said...

You are looking at this bear cycle and comparing it to the last bull cycle. It seems a more interesting data set would be to look at how broad market declines looked in the 2000- 2002 bear cycle. Just sayin'.

Otherwise, nice work.

Tony said...

Also, looking 30 days is not really useful to "traders"; a 5 or 10 day trading day horizon would be more useful. 30 days is for "investors" at this point.

Thanks for your great website.

Brett Steenbarger, Ph.D. said...

Hi Tony,

Yes, for this particular analysis I wanted to see how the current data are similar to or different from what we've seen in recent years. The 30-day horizon would constitute a trading time frame for a hedge fund portfolio manager, but not for active traders, market makers, etc. Thanks for the good observations--


adan said...

possibly, since almost all the 4000+ new-day-low days have been in this extra strong bear market, and there are more stocks now than say in the 1920's and 30's, this is more indicative of a more general pattern within those parameters: # of stocks listed, degree of bear market drop

if the # of stocks listed in the 1920's and 30's could be adjusted for the # listed today, and a comparable # of new lows be adjusted to compare to today's 4000+ level, that might show something re % outcomes

either way, time'll tell us :-)

also, really liked tony's comments re the new investment horizon: 30 days! ;-)