Have you hear of the Titanic Signal?
I hadn't either until this morning, when I read this article from McClellan Financial that had this chart which shows the times (down arrows) in the past 30 years that the number of new 52-week lows exceeds the number of new highs on the NYSE within 7 trading days of a major market high.
Sound confusing? It's not, essentially what it's looking for is a broad sell-off where smaller amounts of stocks are propping up the indexes. That's something we pay attention to all the time in Member Chat (on an intra-day basis), so it's nice to know it makes enough sense to have a name amongst top-notch TA people like McClellan (yes, there are a few I respect). Per their analysis:
The top chart shows all of the instances since 1984 of these preliminary sell signals firing off. You can see that they do tend to cluster around major tops, but they also seem to cry "wolf" a lot at other times when an uptrend continues. Ohama noticed that too, and so he added further criteria to constitute what he called "additional evidence". He wanted to see NL exceed NH for 4 out of 5 days, plus NH declining to less than 1.5% of total issues, and finally to have the DJIA (or SP500) decline for 4 out of 5 days. We now have 2 out of those 3 criteria met, but have not seen the DJIA or SP500 drop for 4 of 5 days.
As we know, we're also getting scary signals from the "Hindenburg Omen," which has some similar criteria to the Titanic Signal and John Bollinger, who created Bollinger Bands, has a good way of describing both of these indications.
Rather than thinking of them as "signals", it is perhaps better to think of them as "alerts". Each can be useful for getting one to pay more attention to bearish signs in other charts and indicators, even as one retains the knowledge that it could turn out not to lead to a big selloff every time.