Tuesday, January 28, 2020

A Different Look at the Hindenburg Omen

I notice that the excellent SentimenTrader site observes that the Hindenburg Omen market pattern was triggered yesterday in the U.S. stock market.  Basically, that pattern occurs when we've had a strong market, but register a high number of stocks making fresh 52-week lows.

As a rule, I'm more than skeptical about technical analysis patterns and especially ones with dramatic titles.  The logic behind the Hindenburg notion, however, makes sense to me.  If we have many stocks making new lows when the overall index is high, that suggests weakness beneath the surface.  A trending market, like a rising tide, should lift all boats.  In the Hindenburg situation, many boats are not lifting and indeed are sinking.

I decided to test out a different version of this pattern.  I collect data on all listed stocks making fresh 3-month new highs and lows on a daily basis.  (Data from Barchart.com).  Going back to the start of my database, late 2010, I examined all occasions in which the SPX was up more than 5% over a prior 3-month basis, but we register more than 500 stocks making fresh 3-month lows.  

Out of over 2200 days in the database, only 46 displayed this pattern.  Interestingly, over the next three trading sessions, we had 18 occasions up, 28 down for an average loss of -.49%.  That compares to an average three-day gain of +.15% for the remainder of the sample.  Even 20 days out, we had 23 up and 23 down for an average loss of -.12%, compared to an average gain of +.92% for the remainder of the sample.

The takeaway here is that context matters.  One of the better predictors of momentum that I've found is not the presence of lots of stocks making new highs, but the absence of new lows.  If no individual segment of the market is weak, it's tough to get overall market weakness going forward.  If, however, a market has been strong, but many of its components have been rolling over, that weakness can spread to the broad indexes.  

As I recently wrote, we're currently dealing with a unique situation in China that is likely to dominate price action going forward.  When we have unique events of this nature, historical models are of limited relevance.  That is why I'm always open to considering historical evidence as hypotheses, but reluctant to embrace it as conclusions.

Further Reading: